2026-11-18 | CS.LG机器学习 | 共 107 篇
[机构]信息由AI分析生成,可能存在错误,仅供参考,以论文实际显示为准
快速导航
1. 深度学习架构与训练方法 13 篇
2. 表示学习、自监督与对比学习 1 篇
3. 强化学习与序列决策 14 篇
4. 生成模型与概率建模 9 篇
5. 优化、泛化与理论分析 5 篇
6. 高效学习、压缩与部署 2 篇
7. 联邦学习、隐私与安全 2 篇
8. 鲁棒性、不确定性与可信学习 5 篇
9. 图学习与结构化数据 1 篇
10. 迁移、元学习与持续学习 5 篇
11. 数据集、基准与评测 5 篇
12. 机器学习应用 6 篇
13. 其他/综合机器学习 39 篇
1. 深度学习架构与训练方法 | 13 篇
1. Regularized Score Centering
正则化得分中心化
AI 总结:该研究针对异步强化学习的不稳定问题,提出正则化得分中心化(RSC)方法,结合反向KL惩罚限制策略不匹配,在六项智能体任务上显著提升样本效率与性能。
链接:https://arxiv.org/abs/2611.21791
机构:UIUC(伊利诺伊大学厄巴纳-香槟分校); CMU(卡内基梅隆大学)
作者:Hao Bai, Chenlu Ye, Xuanchang Zhang, Aviral Kumar, Tong Zhang
英文摘要:Asynchronous reinforcement learning increases training throughput, but data from stale policies and numerical mismatch between training and inference can destabilize learning. To stabilize off-policy learning, score centering (SC) removes the effect of reward baselines on the expected update. Our gradient diagnostics show less extreme gradients with SC than with importance sampling (IS), but SC can still fail when the learner moves far from the behavior policy that generated the data. To limit this policy mismatch, we introduce Regularized Score Centering (RSC), which adds a reverse Kullback-Leibler (KL) penalty to encourage the learner's next-token distribution to stay close to the behavior policy's next-token distribution. RSC works with or without truncated importance sampling (TIS), which caps token importance weights. Our theoretical analysis supports this approach, showing that KL regularization improves convergence guarantees. We evaluate dense and mixture-of-experts models on mathematical reasoning and agentic tasks, including online shopping, tool use, games, and fully asynchronous software engineering. RSC with TIS substantially improves sample efficiency across all agentic benchmarks, attaining higher training rewards with the same number of sampled responses. In final evaluations, it achieves top performance across all six agentic tasks, improving by 14.3% on AppWorld and reaching 9.9 times the strongest baseline's score on BabyAI.
2. The Hidden Cost of MoE Diffusion Decoding: Profiling and Fixing the Sampler
MoE扩散解码的隐藏代价:采样器的分析与修复
AI 总结:该研究分析了MoE扩散解码中logits采样流程的高时间消耗问题,提出两种无损优化方法,结合W4A16专家核实现2.4-2.7倍端到端加速,且精度不变。
链接:https://arxiv.org/abs/2611.21852
机构:Indian Institute of Technology Madras(马德拉斯印度理工学院)
作者:Harshit Samrat, Ramakrishna Paramahamsa Kompella
英文摘要:Masked diffusion language models (dLLMs) decode by iteratively denoising a full canvas of tokens, so every forward pass runs a per-token sampling and stopping pipeline over the entire canvas-by-vocabulary logits tensor. We profile single-user decoding of a 26B-parameter diffusion Mixture-of-Experts model on a bandwidth-bound edge accelerator (NVIDIA GB10) and report a result that is easy to miss: after weight quantisation, this logits pipeline, not the expert GEMMs, is the single largest consumer of decode time. Unlike autoregressive decoding, where sampling touches one position's logits per committed token, diffusion decoding processes S times as many logit entries per output token, where S is the number of denoising steps. We show this pipeline is largely redundant or fusable and remove it in two lossless tiers: a bit-exact tier whose output is byte-identical to the reference decoder, and an accuracy-parity tier using a fused single-pass entropy kernel and Gumbel-max sampling. Combined with a W4A16 expert kernel, the full stack reaches 2.4-2.7 times end-to-end speedup over bf16 at unchanged benchmark accuracy on GSM8K, HumanEval, and MATH, on a single 273 GB/s device. We confirm the finding on a second, independently built diffusion MoE (LLaDA2.0-mini: pipeline 28.8% of decode, fused W4A16 3.3 times at accuracy parity) and show it disappears on a dense diffusion model (3.6%). The cost is specific to MoE diffusion, where quantisation makes the backbone cheap. Code and quantised weights are released.
3. A Log-Timescale Perturbation Bound for Selective State Space Models at Every Sequence Length
任意序列长度下选择性状态空间模型的对数时间尺度扰动边界
AI 总结:本文针对流数据上的Mamba模型,提出对数尺度的选择性状态空间模型扰动边界,无需时间尺度下界,边界紧度优于现有方法且远高于梯度攻击的变化量。
链接:https://arxiv.org/abs/2611.21952
机构:University of Surrey(萨里大学)
作者: Jack Humphries, Daniel Midgley
英文摘要:Mamba models deployed on streaming data, such as limit order book updates, run far past the sequence lengths they were trained on. A guarantee on how much a small change of input can move the hidden state therefore has to hold at every length. The closest existing bound we know of measures the change of the timescale additively, and it grows linearly with the sequence length unless the timescale is bounded away from zero. In this paper, we measure the change on a log scale instead. The key observation is that the decay factor of each step then moves by at most a fixed fraction of that step's input weight, so the per-step errors cannot pile up. Using this, we prove that two runs of a selective-scan cell from a common start stay within a bound that holds at every length, shrinks to zero with the size of the change and needs no lower bound on the timescale. We give a construction showing that, for small changes, this bound is within a factor of about 2 of the best possible, and we extend it to the update used in Mamba's released code when the timescale is bounded above. We then evaluate block-level bounds built from the theorem on two small models trained on one day of order book data. With a bounded timescale, a bound built on the existing analysis is tighter than ours for at most the first 8 steps. With Mamba's softplus timescale, whose certified lower bound is as small as $10^{-21}$, it is looser at every length. Our bounds hold for every input to a block, but they remain more than five orders of magnitude above the changes that gradient attacks find.
4. Weight-Space Programming Doesn't Need Transformers
权重空间编程无需Transformer
AI 总结:该研究针对权重空间编程场景,让前沿LLM设计或生成架构,发现其设计的关联记忆模型比同规模Transformer快17倍、参数少1800倍且准确率更高,表明权重空间编程无需Transformer。
链接:https://arxiv.org/abs/2611.22055
机构:Pohang University of Science and Technology (POSTECH)(浦项科技大学(POSTECH))
作者:Hyunjong Ok, Jaeho Lee
英文摘要:Standard neural architectures such as the Transformer were chosen because they train well with gradient descent and run fast on GPUs. In weight-space programming, a large language model (LLM) writes target weights by hand without gradient-based training, so those reasons no longer apply. Should we keep the Transformer? Given minimal guidance, ten labeled examples, and a sandbox, mainly on multiple-choice questions, we task three frontier LLMs with either generating the weights of a fixed architecture or designing the architecture themselves. When free to design, all three compile plain-text facts into a sparse table queried by a fixed procedure, reaching up to 54.4% test accuracy. An LLM-designed model achieves higher test accuracy than the most accurate released Transformer that the same LLM wrote while using about 1,800x fewer parameters. Without gradients, they choose a table, in which the LLM can state what each weight means and can find and rewrite a wrong one. Each wrong answer points to one row, and removing a row changes about one other answer. From this, we propose an associative-memory-based model up to 17x faster on GPUs than a same-sized Transformer. We also test pretrained weights, other benchmarks, and tasks where models generate answers. Pretrained LLMs remain more accurate; the claim concerns which structure to build when writing weights.
5. BernBRS: Certified Backward Reachability for Deep Bernstein Neural Networks
BernBRS:深度伯恩斯坦神经网络的可验证向后可达性
AI 总结:提出BernBRS框架,实现DeepBern-Nets的可验证向后可达性分析,通过激活模式细化输入空间覆盖,在高维复杂模型上验证了其有效性。
链接:https://arxiv.org/abs/2611.22083
机构:University of California, Irvine(加州大学欧文分校)
作者:Ibrahim Albool, Yasser Shoukry
英文摘要:Ensuring the safety of complex neural network systems requires mathematically rigorous verification. While backward reachability analysis is a critical tool for identifying safe initial states, computing tight, formal bounds remains notoriously difficult to scale. Recent advancements have demonstrated that DeepBern-Nets neural networks utilizing Bernstein polynomial activation functions possess advantageous mathematical properties that enable exceptionally fast and tight forward reachability analysis. Building upon this efficiency, this paper introduces BernBRS, a highly accurate and scalable framework for the backward reachability analysis of DeepBern-Nets. BernBRS simultaneously computes certified inner and outer approximations of the Backward Reachable Set (BRS). Rather than relying on single convex bounding, our method intelligently refines a cover of the input space by analyzing activation patterns, splitting the input domain precisely where an exact decomposition of the relaxation gap predicts the most significant reduction in error. This targeted refinement enables high levels of scalability to highly complex and massive architectures that remain intractable for existing certified methods. We demonstrate the efficacy of our framework on high-dimensional models, including a linearized closed-loop approximation of the Half-Cheetah controller and a distilled Bernstein student of a Transformer block and we empirically study how these model-level guarantees relate to the original Half-Cheetah and TinyBERT classifier, respectively.
6. ReMoE: An Interpretable Mixture-of-Experts Framework for Human Mobility Recovery Forecasting Across Disasters
ReMoE:面向跨灾种人类流动性恢复预测的可解释混合专家框架
AI 总结:ReMoE是面向跨灾种的可解释混合专家框架,通过特征引导路由组合专家模型并结合灾害预测优化结果,在18类灾害数据上实现跨灾种流动性恢复预测,性能优于现有方法且可解释性强。
链接:https://arxiv.org/abs/2611.22133
机构:Shenzhen University of Advanced Technology(深圳理工大学); Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences(中国科学院深圳先进技术研究院); City University of Hong Kong(香港城市大学)
作者:Yilin Lyu, Kang Liu, Wei Wang, Qifan Zhang, Ling Yin, Yushi Bai
英文摘要: Human mobility often declines sharply following disasters and subsequently exhibits complex recovery dynamics. Effective forecasting of this recovery is critical for timely assessment and resource allocation. However, existing approaches are often developed for individual events and operate as black boxes, limiting their transferability and interpretability beyond the events they were trained on. Recent studies have identified recurring post-disaster mobility recovery types across disasters, suggesting that recovery knowledge may be transferable. Motivated by this, we propose ReMoE, an interpretable mixture-of-experts framework for cross-disaster mobility recovery forecasting. ReMoE uses a feature-guided router to adaptively combine specialized experts that learn these heterogeneous yet reusable recovery types, producing a region-specific recovery pattern. It further refines this pattern through residual correction informed by forecasts of future hazard conditions. We evaluate ReMoE on a large-scale mobility dataset covering 18 disaster events spanning hurricanes, floods, winter storms, and wildfires. ReMoE is competitive in-distribution and achieves the strongest performance on unseen events and hazard types, with larger gains under greater distribution shifts. Further analyses show that ReMoE yields interpretable learning outcomes and maintains robust performance under uncertainty. Broadly, ReMoE offers a transferable and traceable forecaster that can support anticipatory post-disaster assessment.
7. Communicate the Change: Efficient Expert Parallelism via Activation Compression
传递变化:通过激活压缩实现高效的专家并行
AI 总结:针对MoE模型专家并行的通信瓶颈,提出EchoMoE架构,通过传递隐藏状态的压缩差值降低通信量,提升吞吐量且保持性能,还支持量化与vLLM集成。
链接:https://arxiv.org/abs/2611.22215
机构:Mila - Quebec AI Institute(Mila - 魁北克人工智能研究所); McGill University(麦吉尔大学); Periodic Labs; CIFAR AI Chair(CIFAR 人工智能席位)
作者:Fengyuan Liu, Jay Gala, Siva Reddy, Dzmitry Bahdanau
英文摘要:Scaling has driven the success of large language models (LLM), with sparse mixture-of-experts (MoE) architectures enabling greater parameter capacity without a proportional increase in computation. At large scales, MoE models rely on expert parallelism to distribute experts across devices. Each MoE layer requires two synchronous collective communication steps during inference: dispatching hidden states to the activated experts and combining the resulting expert outputs. As compute throughput has outpaced interconnect bandwidth, these synchronous collectives have become a major bottleneck to MoE scaling and efficiency. We address this challenge with EchoMoE, a lightweight architecture modification that retrofits existing MoE models, requiring approximately 8,500x fewer training tokens than full pretraining. Motivated by the observation that transformer blocks update the residual stream incrementally, we use a hidden state from an earlier layer as a reference and communicate only the compressed difference between it and the current state, rather than communicating each hidden state in full. EchoMoE periodically refreshes these reference states and transmits only the compressed differences synchronously, reducing the traffic on latency-critical paths without significant computational overhead. EchoMoE reduces inter-node communication volume by 58% and improves throughput by 22.6%, while preserving average performance within 3.8 percentage points across 18 benchmarks. EchoMoE further improves inference throughput by up to 2.38x over bandwidth-constrained Ethernet, paving the way for wide-area distributed inference. EchoMoE also tolerates quantization of the communicated representations without measurable quality loss. We release an efficient implementation for applying EchoMoE to pretrained MoE models, together with an inference integration for vLLM.
8. Spatially local mini-batching for implicit neural representations
隐式神经表示的空间局部小批量处理
AI 总结:该研究提出低秩神经表示(LRNRs)可采用空间局部小批量训练,在6个双曲守恒律问题上,其误差远低于超网络-SIREN、DeepONet等基线模型,且参数仅为同宽度密集网络的十分之一。
链接:https://arxiv.org/abs/2611.22236
机构:Pohang University of Science and Technology (POSTECH)(浦项科技大学(POSTECH)); Washington University in St. Louis(圣路易斯华盛顿大学); Arizona State University(亚利桑那州立大学)
作者:Minseok Choi, Swarup Dhar, Heejae Kwon, Kookjin Lee, Donsub Rim
英文摘要:Neural surrogates based on implicit neural representations (INRs) are typically trained on mini-batches spread over the whole domain, since a generic INR loses accuracy when every point of a batch lies in one small region. We show that low-rank neural representations (LRNRs), INRs whose weight matrices are products of shared bases with coefficients emitted by a small hypernetwork, can be trained with spatially local mini-batches. Each step uses the points of one time snapshot of an adaptive-mesh numerical solution inside one small cell, with snapshots and cells drawn according to the refinement of the mesh. We posit that such training is enabled by the structural property of LRNRs, as a local step reaches the shared bases only through the $r$ coefficients of each layer, and keeps one cell from rewriting the network. By a Kolmogorov-width argument, we show that when restricted to a small cell, the hidden states of LRNRs lie close to a low-dimensional subspace, which suggests the local steps only affect low-rank subspaces of the shared bases. On six hyperbolic-conservation-law problems with solutions computed using adaptive mesh refinement, one local training recipe reaches 1.1-2.3% relative $L^2$ error with a tenth of the parameters of a dense network of the same width. Finally, we provide a comparison by training a hypernetwork-SIREN, a DeepONet, and a Fourier neural operator (FNO) of the same size with the same local batches and training budget. We observe that these baselines reach 16-78% where the LRNR reaches 4.9%. Interestingly, we also observe that local training has a significant effect on the per-cell low-dimensional structure even in a wide generic hypernetwork-SIREN.
9. VRLLM: Turning LLMs into Powerful Tabular Classifiers
VRLLM:将大语言模型(LLMs)转化为强大的表格分类器
AI 总结:Visa研究院通过微调Llama 3.2 1B LLM得到VRLLM,其表格分类性能超越多种表格分类器及该团队的生产级深度学习模型,为紧凑LLMs适配工业表格预测任务提供了工程方案与实证案例。
链接:https://arxiv.org/abs/2611.22308
机构:Visa Research(维萨研究机构)
作者:Yingtong Dou, Shubham Jain, Junpeng Wang, Hadi Abdullah, Vineeth Rakesh Mohan, Yuzhong Chen, Liang Wang, Yiwei Cai
英文摘要: Large Language Models (LLMs) have shown strong performance on text, but whether pretrained LLMs are effective for large-scale tabular classification remains unclear, as many previous works demonstrate that LLMs' tabular data classification performance falls behind tree-based models or neural architectures designed for tabular data. At Visa Research, we managed to fine-tune a pretrained Llama 3.2 1B LLM on billions of rows of labeled tabular data and make it outperform a wide array of tabular classifiers and even our competitive production deep learning model. For ease of reference, we refer to the model as the Visa Research LLM (VRLLM). In this paper, we share how we achieved this exciting result, covering data preparation, training recipe, ablation studies of design choices and deployment validation. We also benchmark frontier LLMs on the same task and analyze their prediction behavior across model families and generations. Our interpretability experiments suggest that VRLLM learns feature-specific predictive associations, while component ablations indicate that the earliest two FFN layers contribute to the emergence of linearly decodable task information. Our paper provides an engineering recipe and a focused empirical case study on adapting compact LLMs to industrial tabular prediction tasks.
10. Scaling Laws for Vision-Language Fusion
视觉-语言融合的缩放定律
AI 总结:本文推导了统一描述视觉-语言融合动态的缩放定律,刻画了早期与后期融合的权衡,证实该定律可指导融合决策。
链接:https://arxiv.org/abs/2611.22351
机构:Noetra Corp.(诺伊特拉公司)
作者:Seng Pei Liew, Teppei Suzuki, Seiya Tokui
英文摘要:Vision-language models (VLMs) have emerged as central foundation models combining multimodal perception with text comprehension. While VLMs are traditionally built via late fusion by continued pre-training an off-the-shelf language model on multimodal (image-text) data, mixing modalities from the start of pre-training (early fusion) has recently gained traction. However, the fundamental trade-offs between these paradigms remain poorly understood, resulting in expensive and heuristic-driven training recipes. In this paper, we systematically analyze the learning dynamics of these multimodal pre-training processes, showing that validation losses during vision-language fusion can be modeled by separating the base learning trajectory, where different data types contribute at distinct rates, from a distribution shift driven by the competition between modalities. We derive a scaling law that unifiedly describes these dynamics, enabling accurate loss predictions across mixture ratios, training steps, and model sizes. Crucially, our law characterizes the trade-off between early and late fusion: under a fixed data budget, early fusion preserves text performance while late fusion favors vision performance, yielding a trade-off that increasingly tilts toward early fusion as model size scales. Extensive experiments confirm that our scaling law reliably captures multimodal learning dynamics and governs fusion decisions.
11. LoopPrune: Recurrent Pruning for Looped Diffusion Transformers
LoopPrune:用于循环扩散Transformer的循环剪枝方法
AI 总结:提出LoopPrune剪枝框架,解决循环扩散Transformer的剪枝问题,在多模型、多基准实验中,30%剪枝后仍保持高准确率,且内存与速度表现优异。
链接:https://arxiv.org/abs/2611.22757
机构:Xiamen University(厦门大学)
作者:Weizhong Huang, Huaijia Kou, Yan Zhang, Fei Chao, Xiawu Zheng, Liujuan Cao
英文摘要:Looped diffusion Transformers reuse shared blocks to deepen computation and match larger models with fewer parameters, reducing storage and deployment burdens. Yet repeated attention and MLP computations remain costly. Pruning reduces storage and computation by removing redundant parameters or units. Training-free methods struggle to resolve recurrent contributions, while training-based methods lack explicit constraints on position-specific errors affecting later inputs and final outputs. We propose LoopPrune, a pruning and recovery framework addressing these requirements. Recurrent importance estimation combines shared-removal effects and position variation to select a shared structure; shared reconstruction pools recurrent input statistics to compensate weights. Loop-position recovery compares complete denoising predictions at identical student states and matches text guidance, updating all student parameters against a fixed dense teacher. Extensive experiments across four open-source models, six benchmarks, and three pruning paradigms demonstrate effectiveness. At 30% structured parameter pruning, Looped-DiT B/16 retains 93.14% of dense six-benchmark average accuracy, exceeding the best baseline by 9.49%. Across PyTorch, vLLM-Omni, and SGLang Diffusion, pruned SDXSv3 reduces peak allocated GPU memory by 34.36-44.64% and accelerates denoising 1.46-1.74x over dense models.
12. Storing the Gemma 4 K=V Cache Once: Exact Global-KV Memory Savings in a Paged Serving Engine with Unchanged Attention Kernels
一次性存储Gemma 4 K=V缓存:在分页服务引擎中实现精确的全局KV内存节省且注意力内核保持不变
AI 总结:该研究针对Gemma 4全局注意力层的KV缓存存储问题,提出仅存储一行640个值的精确方法,在保持注意力内核不变的情况下,使解码KV内存下降1.34-1.40倍,吞吐量提升1.11-1.30倍。
链接:https://arxiv.org/abs/2611.23035
机构:San José State University(圣何塞州立大学)
作者:Kushagra Bainsla
英文摘要:In the global attention layers of Gemma 4 31B, the value is the normalized key projection, and the key is the value scaled by a fixed weight, with rotary position encoding on one quarter of its dimensions. The Gemma 4 report states that this design reduces the global KV cache by up to 37.5%, but the open serving engines that we checked still store a full key and a full value: 1,024 values per KV head and token. We store one 640-value row instead, exact in real arithmetic, and an unchanged paged attention kernel reads it through two overlapping strided views as its key cache and its value cache. In a vLLM prototype with the 16-bit 31B model, the outputs are bit-identical to an emulation of the same arithmetic on the stock cache, 2 of 432 final answers differ from the stock engine (within its run-to-run noise), and the steady decode KV memory per request falls by a factor of 1.34 to 1.40 at 32K to 64K tokens. On a prefill/decode pair of two A100 GPUs with NIXL transfer and CPU offload, a decode engine with a full 15 GiB pool ran 1.38 times the stock decode tokens per step for 16K-token prompts (repeated to 0.06%) and 1.24 times for 8K-token prompts; a model of the vLLM block pool predicted both ratios to within 1% (the engine was 0.7% to 0.9% higher) and matched the change between them. The eager decode throughput rose by 1.30 and 1.11 times, and on this path the compact engine gave the same output tokens as the emulation for 8 prompts at batch 1. The run-time prototype makes an eager step 4% to 9% slower at equal batch, and the method does not yet stack with an 8-bit KV cache at the default scale.
13. Large-scale Neuron Community Discovery Reveals Neuron-Sparse Behaviors in Language Models
大规模神经元社区发现揭示语言模型中的神经元稀疏行为
AI 总结:本研究提出神经元社区发现方法,在 Llama-3.1-8B 上发现支持多种功能的神经元社区,证实小神经元组可作为跨提示复用的计算组件,且能大规模发现。
链接:https://arxiv.org/abs/2611.23112
机构:Goodfire(古德法尔(Goodfire)); Imperial College London(伦敦帝国学院)
作者:Igor Shilov, Thomas Fel, Tal Haklay, Lucius Bushnaq, Owen Lewis
英文摘要:What reusable units of computation can we discover within a language model's neurons? Prior work has shown that individual forward passes can often be sparsified post-hoc to small groups of neurons, but it has been unclear whether these groups form coherent computational entities reused across inputs. We introduce neuron community discovery, an unsupervised method for finding neuron groups that recur across prompts and are causally necessary and sufficient, within the layers they span, for common computations. The method combines gradient-based attribution, a graph connecting neurons that are repeatedly attributed together, and community detection. Applied to Llama-3.1-8B using ten million prompts, it discovers thousands of candidate communities. We evaluate them on held-out prompts through three causal tests: community necessity, sufficiency within a layer subset, and layer necessity. These tests identify groups whose removal disrupts predictions and whose retention preserves predictions despite ablating the remaining MLP neurons in the tested layers. Discovered communities support functions like grammatical agreement, token-probability calibration, fixed phrases, in-context copying, addition, and multiplication. With communities capped at 128 neurons across all layers, 10.3% of held-out prompts have at least one layer's contribution explained on a neuron-sparse basis. Coverage remains 8.7% when requiring at least four layers and 3.3% when requiring at least twelve. These results show that small groups of existing neurons can provide reusable computational components across prompts, and that such groups can be discovered at scale.
2. 表示学习、自监督与对比学习 | 1 篇
14. Urban Representation Learning for Identifying Potential Disparities in Crime Reporting
用于识别犯罪报告潜在差异的城市表征学习
AI 总结:提出自监督城市表征学习框架,结合NPD量化区域犯罪报告差异,在墨西哥城1058个H3区域的180万条犯罪记录实验中,其表征性能优于Hex2Vec,NPD排名稳定,可用于识别需进一步调查的区域。
链接:https://arxiv.org/abs/2611.21825
机构:School of Engineering and Sciences(工程与科学学院); Tecnologico de Monterrey(蒙特雷理工学院)
作者:Jorge Valverde-Rebaza, Carlos Paez De La Cruz, Pedro Pérez-Collado, Daniel Sánchez-Cataño, Eashley Martínez-Vergara
英文摘要:Open crime data support urban safety analysis but reflect reported incidents rather than the complete distribution of criminal activity, potentially introducing spatial distortions in downstream analysis. We propose a self-supervised urban representation learning framework for identifying unusual reporting patterns relative to representation-similar peers using only georeferenced crime records. The method constructs complementary hourly, weekly, and static reporting profiles, learns scale-specific representations under geographic self-supervision, and integrates them into a unified latent space. Neighbor Profile Divergence (NPD) is then used to quantify disagreement between each region's reporting profile and those of its retrieved peers. Experiments on more than 1.8 million crime records from Mexico City, aggregated into 1,058 H3 regions, show that the proposed representation achieves a mean Silhouette score of 0.4052, compared with 0.2863 for Hex2Vec. NPD rankings remain highly stable across neighborhood sizes, with Spearman correlations from 0.9720 to 0.9951, while most high-divergence candidates persist across configurations. The retrieved peers also extend substantially beyond immediate geographic neighborhoods, indicating that the learned representation captures structure beyond local spatial proximity. High-divergence regions exhibit spatially organized patterns and, in several cases, differences in incident volume relative to peer expectations. These results support NPD as an internal auditing and prioritization mechanism for identifying regions that warrant further investigation, rather than as direct evidence of underreporting.
3. 强化学习与序列决策 | 14 篇
15. Adaptive Memory for Online Reinforcement Learning under Sparse Model Changes
稀疏模型变化下在线强化学习的自适应记忆
AI 总结:针对稀疏模型变化下的在线强化学习,提出自适应记忆算法,通过逐次检查历史数据复用情况,仅需最近O(1/ε)次部署观测值,实现最优回报与策略回报的小平均差距。
链接:https://arxiv.org/abs/2611.21879
机构:North Carolina State University(北卡罗来纳州立大学); RCC Institute of Information Technology(RCC信息技术学院); Columbia University(哥伦比亚大学)
作者:Sreejeet Maity, Sukanya Modak, Leonardo F. Toso
英文摘要: We study the discounted reinforcement learning (RL) problem in a setting where the underlying environment changes between deployments. Each deployment provides $H$ observations from a fixed MDP. The reward function and transition kernel are fixed within each deployment. Across deployments, at most an $\epsilon >0$ fraction of the reward and transition kernel entries may change. In this setting, learning each model to deploy from scratch may discard useful observations from past deployments, whereas pooling all past observations can bias the estimates of entries that have already changed. We then propose an algorithm that for each state and action pair, selects how much history to reuse to estimate the reward and probabilities of the next state. We start with observations from the current deployment and add earlier deployment observations one at a time, the most recent first. After each addition, we check whether a single average reward agrees with the estimates from all histories considered until that point. We apply the same check to the distribution of next states. For each estimate, we leverage the last history that passes these aforementioned checks. The optimal control policy is then computed on top of the fitted MDP. We characterize the average gap between the optimal return and the return of the learner's policies over a finite number of deployments. We demonstrate that when $0< |S|\epsilon \leq 2$, with $|S|$ being the cardinality of the state space, our algorithm needs only the observations from the most recent $O(1/\epsilon)$ deployments, and achieves a long-run average gap of at most $\widetilde{O}(\sqrt{\epsilon/H})$.
16. Cancellation-Free Constrained SquareCB for Safe Controller Selection
无取消约束SquareCB算法用于安全控制器选择
AI 总结:该研究提出无取消约束SquareCB算法,用于约束上下文博弈的安全控制器选择,在两类基准任务中违规程度显著低于现有方法,遗憾表现更优。
链接:https://arxiv.org/abs/2611.21880
作者:Kartik Pandit, Sourav Ganguly, Arnob Ghosh
英文摘要:We study constrained contextual bandits, in which a learner chooses one of $K$ actions in each round to minimize loss while keeping the expected cost of its play below a budget. Existing reductions to online regression weight the cost by a Lagrange multiplier learned from past constraint residuals, so slack in some rounds can offset violations in others. We instead run SquareCB proposed in \citep{foster2020beyond} with a rectified weight on the cost: it is zero when the estimated cost is within the budget, and otherwise just large enough to bring it back to the budget. Under realizability and Slater's condition with margin $\delta$, the algorithm attains regret $O(\sqrt{KT\,\mathrm{Reg}_{\mathsf{Sq}}(T)}/\delta)$ and cancellation-free violation $\sum_t(V_c^{p_t}-b)^+=O(\sqrt{KT\,\mathrm{Reg}_{\mathsf{Sq}}(T)})$, where $\mathrm{Reg}_{\mathsf{Sq}}$ is the error of the regression oracle. The same bounds hold when the budget must be met in every context, potentially adversarially chosen contexts. In controller selection for a safe-control-gym quadrotor and a BOPTEST building, the violation is 6 to 9 times lower than that of primal-dual and virtual-queue baselines, with a lower or comparable regret.
17. World-Grounded Reinforcement Learning for Pollution-Aware Bird Movement Modeling Using a Multi-Instance DQN Framework
基于多实例DQN框架的面向污染感知鸟类运动建模的世界接地强化学习
AI 总结:该研究提出多实例DQN框架,以世界反馈(AQI)为训练信号,结合Xeno-canto数据集等,构建污染感知的鸟类运动模型,其DQN方法在多基准对比中表现更优,为生态决策建模提供可扩展基础。
链接:https://arxiv.org/abs/2611.21895
机构:Vellore Institute of Technology (VIT)(韦洛尔理工学院); MBZUAI(穆罕默德·本·扎耶德人工智能大学)
作者:Shivin Mangal, Divyansh Chawla, Anshu Garg, Isshaan Singh
英文摘要:Birds are modeled as environmentally sensitive agents whose movement reflects real-world air quality conditions, where AQI serves as a measurable world feedback signal representing pollution-driven consequences. Instead of relying on human preference or synthetic rewards, the system uses environmental outcomes as the primary training signal, enabling policy learning grounded in ecological impact. We propose a multi-instance reinforcement learning simulation in which agents navigate a 2D grid representation of India integrated with a precomputed AQI matrix and geographic state boundaries. The framework combines bird call metadata from the Xeno-canto dataset, spatial discretization, and real environmental data to learn adaptive navigation strategies under pollution constraints. To ensure scalability and efficiency, repeated real-time API queries are replaced by a structured AQI mapping that enables fast lookup during training and execution. Multiple baselines, including Monte Carlo, Q-Learning, and Value Iteration, are evaluated for comparison. Deep Q-Network demonstrates stronger stability, adaptability, and cumulative performance across varying environmental conditions, establishing a scalable foundation for world-feedback driven ecological decision modeling.
18. GP--RL--MPC: Uncertainty-Aware Reinforcement Learning-Augmented Model Predictive Control with Gaussian Process Postdecision State-Value Functions
GP--RL--MPC:结合高斯过程决策后状态值函数的不确定性感知强化学习增强型模型预测控制
AI 总结:提出GP--RL--MPC方法,结合GPR与OAPI学习评论家,经CSTR实验使SSE较基线MPC平均降低80.52%。
链接:https://arxiv.org/abs/2611.22038
机构:University of Southern California(南加州大学)
作者:Jon Li, Jay H. Lee
英文摘要:We propose Gaussian Process Reinforcement Learning-Augmented Model Predictive Control (GP--RL--MPC), which uses a learned critic as the MPC terminal penalty to account for costs beyond the prediction horizon. Gaussian process regression (GPR) enables data-efficient critic learning from limited, noisy rollouts within optimistic approximate policy iteration (OAPI), with soft updates mitigating policy chattering. A tunable risk coefficient weights GP epistemic uncertainty in the terminal cost, regulating sensitivity to uncertain critic predictions. The postdecision state formulation shifts disturbance averaging offline to account for aleatoric uncertainty from the terminal transition onward. Under matching objectives and admissible actions, a one-step horizon recovers greedy value-space RL. After fitting GP hyperparameters by exact marginal likelihood, we construct a sparse deterministic training conditional (DTC) posterior using inducing points selected by pivoted Cholesky (PC), reducing online evaluation costs as the training dataset grows. In a continuous stirred-tank reactor (CSTR) subject to exogenous disturbances, GP--RL--MPC with an appropriately selected risk coefficient and policy screening reduces the sum of squared errors (SSE) by an average of 80.52% relative to an otherwise matched baseline MPC across five random seeds (sample standard deviation: 3.48%).
19. RiDEx: Reward-Informed Directional Exploration for Diffusion Reinforcement Learning
RiDEx:用于扩散强化学习的奖励感知方向探索
AI 总结: 提出RiDEx方法将扩散强化学习的探索集中于奖励相关子空间,在Stable Diffusion 3.5 Medium上的四个任务中获最高峰值奖励,且目标达成速度较DiffusionNFT提升15至29倍。
链接:https://arxiv.org/abs/2611.22052
机构:University of Science and Technology of China(中国科学技术大学); Li Auto Inc.(理想汽车); Huazhong University of Science and Technology(华中科技大学); Tongyi Lab(通义实验室); Xi’an High-tech Research Institute(西安高新技术研究院)
作者:Song Yan, Chenfeng Wang, Wei Zhai, Xingchen Wang, Tao Zhang, Yawen Shao, Shi Pan, Min Li, Zheng-Jun Zha
英文摘要:Reward-based diffusion post-training updates model parameters from task rewards. While existing approaches improve sampling and policy optimization, they still rely largely on isotropic random perturbations in a high-dimensional latent space for exploration. Such variation often spends the perturbation budget on changes with little influence on the target reward, making exploration inefficient. We study where diffusion models should explore and how to learn from that exploration. We introduce Reward-Informed Directional Exploration (RiDEx), which concentrates local exploration in reward-relevant subspaces of the denoising velocity field. Conditional residuals or differentiable reward gradients propose the subspace, and equal-and-opposite perturbations at the same generation state measure its terminal reward response. The reward difference supplies signed local supervision in the explored coordinates, while all generated images contribute to global DiffusionNFT learning. RiDEx thereby couples exploration-space selection with feedback about the changes it produces. On Stable Diffusion 3.5 Medium, RiDEx attains the highest peak rewards among the evaluated methods on four alignment and preference tasks. Its gradient-direction instantiation achieves 15x and 29x target-attainment speedups over DiffusionNFT on HPSv2 and PickScore, measured in H200 GPU hours. Controlled experiments confirm that reward-aligned directions elicit stronger responses at equal perturbation magnitude, and that applying their measured credit improves curve-average reward across all four tasks.
20. When does deep reinforcement learning pay off in HVAC supervisory control? Four small Modelica case studies trained directly on FMUs
深度强化学习在HVAC监督控制中何时能发挥作用?四个直接在FMU上训练的小型Modelica案例研究
AI 总结:该研究通过四个Modelica案例,对比直接在FMU训练的DRL智能体与规则控制器、近最优参考,发现DRL在带冷水存储的HVAC控制中收益显著,无存储场景收益小,预冷场景不如手动规则,还指出了FMU训练的实际问题。
链接:https://arxiv.org/abs/2611.22144
机构:Queensland University of Technology(昆士兰科技大学); International Laboratory for Air Quality and Health, Queensland University of Technology(昆士兰科技大学国际空气质量与健康实验室)
作者:Bingnan Zhao, Harshala Gammulle, Tharindu Fernando
英文摘要:Deep reinforcement learning (DRL) is often proposed for the supervisory control of heating, ventilation and air-conditioning (HVAC) systems. The reported gains are, however, difficult to interpret, because the baselines are often weak and the attainable optimum is seldom known. This paper examines four small, reproducible case studies based on an office building in Brisbane, Australia: pre-cooling under a time-of-use tariff; chiller staging with condenser water reset; chilled water supply temperature reset with a wet cooling coil; and chilled water storage under time-of-use prices and a demand charge. The plants were modelled with the Modelica Buildings library, and in three of the cases a soft actor-critic agent was trained directly on the exported functional mock-up unit (FMU). Each agent was compared with tuned rule-based controllers and, where practicable, with a near-optimal reference from hourly exhaustive search or dynamic programming with perfect foresight. In the two cases without storage the agent reached the near-optimal reference, but the margin above a well-chosen rule was small: 1.3% for condenser water reset and 0.18% for chilled water reset. In the latter case two common reset rules consumed more energy than a fixed set point. With a chilled water storage tank the margin was considerably larger. The agent outperformed both rules (-7.8%, compared with -3.9% and -6.3%) and adopted midday charging during the solar off-peak period, yet it obtained only about 45% of the saving achieved by the perfect-foresight schedule and did little to limit the demand peak. In the pre-cooling case a hand-tuned rule remained superior (-12.0% compared with -5.6%). These results suggest that studies of DRL for HVAC control should state the margin between the best rule and a near-optimal reference. Some practical difficulties encountered in training on co-simulation FMUs are also reported.
21. What RL on Expert-Built Tasks Teaches Nemotron 3.5 Super
基于专家构建任务的强化学习对Nemotron 3.5 Super的启发
AI 总结:该研究通过在五个专家构建任务数据集上用RL后训练Nemotron 3.5 Super的LoRA适配器,在12个基准测试中取得显著性能提升,且各适配器针对不同任务弥补了基础模型的遗漏。
链接:https://arxiv.org/abs/2611.22161
机构:Surge AI
作者:Sushant Mehta, Edwin Chen
英文摘要:We post-train NVIDIA's Nemotron 3.5 Super (120B total, 12B active parameters) with reinforcement learning (RL) on five expert-built task datasets, one at a time: real-world software engineering, complex constraints, long-horizon tool use, professional chart understanding and enterprise document reasoning. Each run trains a fresh rank-32 LoRA adapter with CISPO on Tinker, with no supervised stage, and each adapter is compared with the base model on the same tasks, harness and settings. On twelve headline benchmarks, eight built by other groups and four by Surge AI, the gains range from +4.5 points on IFBench to +13.8 on GDP_pdf, which rises from 10.4% to 24.2%; SWE-bench Pro V2 rises from 86.9% to 92.1%, Terminal-Bench 2.1 from 63.9% to 70.3%, MultiChallenge from 58.3% to 66.6%, AutomationBench from 14.8% to 27.5% and Chartography from 39.4% to 50.6%. The document adapter, trained without API tools, also gains +8.0 points on AutomationBench. In attempts read side by side with the base model's on tasks each adapter newly solves, the base model's failures recur in a few modes, and each adapter does more often what the base model missed: the tool-use adapter finds the data and the rules before acting, the coding adapter writes its own test or scratch files on SWE-bench Pro V2, the constraints adapter carries earlier turns into its answers, the chart adapter measures the marks in code, and the document adapter reads AutomationBench's spreadsheets.
22. Towards Optimism-Pessimism Trade-off in Model-based Offline-to-Online Reinforcement Learning
面向基于模型的离线到在线强化学习中的乐观-悲观权衡
AI 总结:该研究针对基于模型的离线到在线强化学习中离线与在线的分布偏移问题,提出双目标公式生成帕累托策略池,结合 MOSAIC 与上下文多臂老虎机算法,构建 HiP3 框架,提升了不同任务下的综合性能。
链接:https://arxiv.org/abs/2611.22172
机构:City University of Hong Kong(香港城市大学); Tsinghua University(清华大学); Jiangxi University of Finance and Economics(江西财经大学); University of Connecticut(康涅狄格大学); Shanghai University of Finance and Economics(上海财经大学); Sun Yat-sen University(中山大学); Southern University of Science and Technology(南方科技大学)
作者:Guochen Zhou, Yijun Yang, Qiqi Duan, Qing Su, Weiming Ou, Li Shen, Shihao Ji, Yuhui Shi, Zhisong Zhang
英文摘要:Model-based offline-to-online reinforcement learning (RL) enables sample-efficient adaptation by leveraging offline pre-training for online fine-tuning. However, the distribution shifts between offline and online stages often hinder fine-tuning performance. Many existing methods encode the optimism-pessimism trade-off through a single-objective formulation. Selecting its trade-off coefficient through online fine-tuning can require costly bi-level optimization. We identify the balance established during offline training as a key challenge: in our motivating experiments, optimistic policies adapt better to novel tasks, whereas pessimistic policies perform better on the same online tasks. To address this challenge, we propose a bi-objective formulation that captures this trade-off, yielding a Pareto policy pool during offline training. This policy pool enables adaptive selection during online fine-tuning across diverse online tasks. To generate the pool, we introduce Multi-Objective Soft Actor-critIC (MOSAIC), which solves bi-objective problems and constructs diverse Pareto policies. After offline training, a contextual bandit algorithm hierarchically selects the most suitable policy for fine-tuning at each online interaction step. Empirically, our pipeline, Hierarchical Pareto Policy Pool (HiP3), achieves the strongest aggregate performance across both same and novel online tasks and maintains its advantage under stronger distribution shifts. Comprehensive ablations further clarify the roles of the policy pool, MOSAIC-based pool construction, and the online selection mechanism.
23. PCGRL for Complex Platformers: Synthesizing Playable and Diverse Levels via Inflated Envelopes and Forward Curricula
用于复杂平台游戏的PCGRL:通过膨胀包络与前向课程合成可玩且多样的关卡
AI 总结:提出物理感知、课程引导的PCGRL框架,通过膨胀体积包络、多阶段前向课程与高熵随机推理,合成可玩且多样的复杂平台游戏关卡,性能优于传统基线。
链接:https://arxiv.org/abs/2611.22250
机构:The University of Hong Kong(香港大学); Harbin Institute of Technology(哈尔滨工业大学)
作者:Di Wu
英文摘要:Procedural Content Generation via Reinforcement Learning (PCGRL) has emerged as a promising paradigm for automated game development. However, existing PCGRL frameworks still suffer from severe exploration bottlenecks and scaling limitations when faced with long-sequence generation tasks governed by strict rigid-body rules and non-linear kinematic boundaries (e.g., platformer constraints). Consequently, traditional approaches frequently fail to maintain reliable level solvability on large-scale canvases. To address these challenges, we present a physics-aware, curriculum-guided PCGRL framework. First, we construct an "Inflated Volume Envelope" mechanism, which provides a physics-constrained representation that incorporates continuous kinematic feasibility into the discrete state space of the MDP. Second, to alleviate the sparse reward challenge induced by extremely long decision sequences, we introduce intermediate state-transition feedback and a multi-stage forward extension curriculum learning framework, enabling informative value estimation and stable credit assignment over long-horizon trajectories. Finally, to resolve the issue where deterministic policies after convergence generate highly monotonous map paths with a lack of diversity, we activate high-entropy stochastic inference during the synthesis phase, regularized by a post-hoc parallel rejection filtering mechanism. Experimental results demonstrate that our framework significantly enhances structural diversity while maintaining high physical playability across multi-scale environments. Comparative evaluations across multi-scale canvases demonstrate that our framework maintains superior convergence stability and dimensional scalability, reliably synthesizing solvable levels where traditional PCGRL baselines degrade.
24. Eyes in the Update? Learning Where Rewards Land in Multimodal Reasoning
更新中的“眼睛”?学习多模态推理中奖励的落点
AI 总结:本研究提出 Gaze-RL 方法,无需额外监督或架构修改,通过模型自身多模态计算推断奖励分配,提升 Qwen2.5-VL 与 InternVL3.5 在图像视频基准的性能。
链接:https://arxiv.org/abs/2611.22438
作者:Lixuan He, Tianzhe Chu, Shulin Tian, Shikang Zheng, Feng Chen, Yi Ma
英文摘要:Multimodal post-training has moved from broad instruction tuning toward reinforcement learning for image and video reasoning, where outcome rewards can improve final-answer accuracy. Yet these rewards are assigned to complete generated trajectories, while visual evidence is used only at sparse and uneven positions inside the response. Recent work has begun to make multimodal RL more token-sensitive through grounding signals or external token-level weighting. We ask a more specific question: can the allocation of the same trajectory-level reward be inferred from the model's own multimodal computation, rather than from an auxiliary annotator, perception module, or process label? We propose Gaze-RL, a representation-driven policy optimization method that redistributes completion-level advantage through a latent reward landing field. The field is computed from internal cross-modal geometry and local predictive uncertainty, assigning larger update weight to positions that are both visually anchored and decision-bearing while remaining mean-one, so the total credit of each rollout is preserved and only its positionwise allocation changes. Gaze-RL requires no process supervision, auxiliary perception module, or architectural modification, and improves both Qwen2.5-VL and InternVL3.5 across a broad range of image and video benchmarks.
25. ReMaRL: Learning to Unmask and Remask in Frozen Diffusion Language Models
ReMaRL:在冻结的扩散语言模型中学习解掩蔽与重掩蔽
AI 总结: ReMaRL 是一种轻量级强化学习策略,用于在冻结的 LLaDA-8B-Instruct 扩散语言模型中优化解掩蔽、保留或重掩蔽决策,在指定推理预算下的数学与代码任务中表现优于其他策略。
链接:https://arxiv.org/abs/2611.22576
机构:JetBrains Research(JetBrains研究院); Constructor University(康斯特大学)
作者:Egor Shibaev, Danila Biktimirov, Ivan Klimov, Artem Makoian, Viacheslav Meshchaninov, Dmitry Vetrov
英文摘要:Masked diffusion language models generate tokens in parallel, and the order in which tokens are unmasked affects final-answer accuracy. Several methods learn unmasking policies from task rewards, but these policies cannot remask earlier predictions. Remasking lets the model correct those predictions using newly generated context. Heuristic rules guide both unmasking and remasking, while supervised remaskers learn to detect token errors. Neither approach directly optimizes these decisions for final-answer correctness. To address this gap, we introduce ReMaRL (Remasking with Reinforcement Learning), a lightweight policy that learns which tokens to unmask, retain, or remask. We train this roughly 0.3M-parameter policy with Group Relative Policy Optimization (GRPO) from final-answer rewards while keeping the diffusion model frozen. With LLaDA-8B-Instruct, ReMaRL achieves the highest scores among the compared unmasking and remasking policies within the tested inference budgets of 4-32 model calls on math and 8-16 calls on code.
26. WATERFALL: Workflow for Adaptive Training with Evolutionary Reward Formulation and Automated Learning Loops
WATERFALL:结合进化奖励公式与自动学习循环的自适应训练工作流
AI 总结:WATERFALL是一种用于POMDP环境的自动奖励发现迭代工作流,通过角色条件生成、视觉语言模型评判的进化锦标赛及纵向评估优化奖励程序,在POMDP场景中优于相关基线。
链接:https://arxiv.org/abs/2611.22854
机构:ETH Zurich(苏黎世联邦理工学院); UCL(伦敦大学学院)
作者:Eleftherios Triantafyllidis, Filippos Christianos, Zhibin Li, Bernd Bickel
英文摘要:Reward design remains a central bottleneck in reinforcement learning (RL), particularly in sparse-reward, long-horizon and partially observable settings requiring dependency chaining, memory, and deceptive affordance disambiguation. While recent foundation-model approaches reduce manual reward engineering, most assume goal-conditioning, task descriptions, expert demonstrations, or ground-truth metrics. These assumptions are problematic in partially observable environments, where task-relevant objectives, affordances, and dynamics must be discovered through interaction rather than disclosed a priori. We formalise this problem setting by introducing WATERFALL, an iterative population-based workflow for automated reward discovery under the strict observability constraints inherent in POMDP environments. WATERFALL achieves this by (i) synthesising diverse programmatic reward candidates via persona-conditioned generation, (ii) evaluating candidate behaviour from raw visual rollouts utilising a Swiss-system evolutionary tournament judged by vision-language models to identify elites, and (iii) leveraging longitudinal assessment histories to iteratively mutate, escalate, or prune candidates. We evaluate WATERFALL across fully observable and reveal-gated partially observable environments ranging from discrete to continuous, comparing against sparse-reward RL, intrinsic-motivation and foundation-model reward-synthesis baselines. Our results indicate that WATERFALL discovers and refines reward programs without access to privileged information, outperforming baselines that retain their native information advantages, most clearly in the POMDP regime. Extensive ablations substantiate that visual evidence, tournament-based selection, and iterative refinement are jointly important. Enforcing this contract is what separates autonomous reward discovery from benchmark-privileged reward shaping.
27. Action-outcome coupling: a pre-investment feasibility audit for offline reinforcement learning in supply chain operations
行动-结果耦合:供应链运营中离线强化学习的预投资可行性审计
AI 总结:该研究针对供应链运营中离线强化学习预投资可行性问题,提出行动-结果耦合指标与四级审计方法,验证了其在电商日志上的有效性及在部分物流日志中的失败情况。
链接:https://arxiv.org/abs/2611.22921
作者:Tung Dang, The Hung Phung, Son Lam Nguyen, Tu Nguyen
英文摘要:Prescriptive analytics initiatives in logistics, such as offline reinforcement learning and next-best-action systems, are often funded before anyone verifies that the available historical logs can support an improved decision policy. We formalise this pre-investment feasibility question via action-outcome coupling: a total-effect variance index measuring how much expected outcomes change with the logged action. Coupling provides an algorithm-independent upper bound on any policy's achievable value, expressed in monetary units per decision for both single-step and discounted sequential objectives. We also analyse the asymptotic false-licensing risk of an invalid workflow: fitting a value model and evaluating the model's greedy policy on the same data. The false-licence rate is driven mainly by the number of state cells over which selection is performed, rising from the nominal test level with one cell to near certainty with a few dozen, while overlap has a smaller effect. We then propose an audit that assigns logs to four tiers of decision support, separately for fixed and contextual policies, and returns an explicit stopping rationale. On a randomised e-commerce log, the audit endorses a fixed recommendation policy, certifying gains equal to 30 per cent of the theoretical ceiling; the remaining gap reflects bound looseness under an 80-action design. Two public logistics logs fail at different checkpoints: one because the target is an accounting identity of the action, and another because, with 27 clusters, estimated coupling stays below the design's detectability threshold even though a conservative offline learner certifies an improvement in 29 of 30 seeds.
28. Learning to Undo: Transfer Reinforcement Learning under State Space Transformations
学习撤销:状态空间变换下的迁移强化学习
AI 总结:本研究提出一种通过匹配状态特征统计量学习撤销映射的算法,可在状态空间变换的MDP间迁移策略,理论与实验均表明其样本效率优于从零开始的强化学习。
链接:https://arxiv.org/abs/2611.23081
机构:Boston University(波士顿大学); Broad Institute of MIT and Harvard(麻省理工学院与哈佛大学博德研究所)
作者:Mridul Mahajan, Aldo Pacchiano, Xuezhou Zhang
英文摘要:Transfer learning in reinforcement learning (RL) has shown strong empirical success. In this work, we take a more principled perspective by studying when and how transferring knowledge between MDPs (a source and a target) can be provably beneficial. Specifically, we consider the case where there exists an undo map such that applying this map to the target's state space recovers the source exactly. We propose an algorithm that learns this map by matching state feature statistics gathered from both MDPs, and then uses it to transfer the source policy to the target. We theoretically justify the algorithm by analyzing the setting when the undo map is linear and the source is linearly-$Q^\star$ realizable, where our approach has strictly better sample complexity than tabula rasa RL in the target MDP. Empirically, we demonstrate that these benefits extend beyond this regime: on challenging continuous control tasks and Atari games, our method achieves significantly better sample efficiency. Overall, our results highlight how shared structure between tasks can be leveraged for efficient transfer of policies across environments.
4. 生成模型与概率建模 | 9 篇
29. FACE: Flow-Guided Angular Channel Estimation Framework for mmWave MIMO Systems
FACE:面向毫米波MIMO系统的流引导角度信道估计框架
AI 总结:针对毫米波MIMO系统信道估计的高维、导频有限等挑战,提出基于流匹配的FACE框架,其精度优于传统及扩散式估计器,导频需求少、计算成本低,性能接近完美CSI基准。
链接:https://arxiv.org/abs/2611.21698
机构:Centre for Wireless Communications (CWC), University of Oulu(奥卢大学无线通信中心)
作者:Samitha Gunarathne, Eslam Eldeeb, Nurul Huda Mahmood, Matti Latva-aho
英文摘要:Acquiring accurate channel state information (CSI) is essential for millimeter-wave (mmWave) multi-input multi-output (MIMO) systems. However, it remains challenging due to high-dimensional channels, limited pilot observations, and measurement noise. To address these challenges, we propose flow-guided angular channel estimation (FACE), a novel channel estimation framework based on flow matching, an emerging generative modeling paradigm that learns continuous transport dynamics between probability distributions. FACE exploits the inherent angular-domain sparsity of mmWave channels and incorporates the learned flow matching model as a generative channel prior. Numerical results show that FACE consistently achieves higher channel estimation accuracy than both conventional estimators and diffusion-based generative estimator across diverse channel and system configurations. Moreover, FACE maintains high estimation accuracy with fewer pilot observations and achieves an average achievable rate close to the perfect-CSI benchmark. Compared with diffusion-based estimation, FACE requires substantially fewer sampling steps while achieving superior or comparable estimation accuracy, reducing inference time and computational cost. Overall, the proposed framework demonstrates the potential of flow matching for accurate, pilot-efficient, and computationally efficient mmWave MIMO channel estimation.
30. Learning Calibrated and Samplable Predictive Distributions of Surface-Ocean $Δp$CO$_2$ Using Tabular Diffusion Models
使用表格扩散模型学习海表ΔpCO₂的校准且可采样的预测分布
AI 总结:本研究用Treeffuser条件扩散模型学习海表ΔpCO₂的预测分布,在SOCAT测试集上取得最优不确定性量化结果,同时保留与XGBoost相当的点预测技能。
链接:https://arxiv.org/abs/2611.21769
机构:CUNY New York City College of Technology(纽约城市大学纽约市技术学院); Lamont-Doherty Earth Observatory, Columbia University(哥伦比亚大学拉蒙特-多尔蒂地球观测站)
作者:Gabriele Accarino, Mohammad M. Amirian, Thea H. Heimdal, Amanda R. Fay, Galen A. McKinley, Viviana Acquaviva
英文摘要:The ocean absorbs almost 30% of anthropogenic CO$_2$ emissions, but quantifying this sink requires estimating surface-ocean CO$_2$ from observations that are sparse in space and time. Most existing machine-learning estimates of this field are deterministic, giving a single value per grid point, with uncertainty approximated post hoc via residuals, ensembles, or prescribed distributions. We instead learn a conditional predictive distribution for each observation: using Treeffuser, a conditional diffusion model with gradient-boosted-tree score functions, we estimate the full predictive distribution of $\Delta p$CO$_2$ (the sea-air $p$CO$_2$ difference) from 13 environmental, spatial, and temporal predictors. On a held-out SOCAT test set, Treeffuser attains the lowest continuous ranked probability score among the uncertainty-quantification methods evaluated (8.229 $\mu$atm) and the coverage closest to the nominal 68% / 95% levels (0.686 / 0.930), while retaining point-prediction skill comparable to a deterministic XGBoost baseline (RMSE 18.4 vs. 17.7 $\mu$atm; $R^2$ 0.77 vs. 0.79). These results show that conditional diffusion can provide well-calibrated, observation-level uncertainties while largely preserving point-prediction skill. Because Treeffuser is generative, its samples can support probabilistic analyses of $\Delta p$CO$_2$; extending this to spatially coherent field reconstruction will require modeling dependence across grid cells and evaluation on spatially held-out data.
31. DSADiff: Data Structure-Aware Diffusion for Crystal Structure Prediction via Mitigating Complexity- and Imbalance-Induced Biases
DSADiff:通过缓解复杂度与不平衡诱导偏差实现晶体结构预测的数据结构感知扩散模型
AI 总结:本研究针对现有扩散型晶体结构预测模型的生成偏差问题,提出LAS框架分析偏差来源,进而构建DSADiff模型,在三个基准数据集上实现匹配率最大提升9%,取得新的最先进性能。
链接:https://arxiv.org/abs/2611.22050
作者:Jiang Xiaobo, Dinghong Lai
英文摘要: Crystal Structure Prediction (CSP) is an important step in accelerating the generation and discovery of new materials. In recent years, diffusion-based CSP methods have made significant progress. However, existing models still suffer from generation bias when facing diverse materials spaces, while systematic approaches for tracing the sources of such performance variations remain limited. To address this issue, we propose LAS (Layered Analysis Strategy), a mechanism analysis framework that identifies crystal complexity and elemental composition imbalance as two key factors affecting generation difficulty and further reveals their relationships with model generation performance. Building on these findings, we further propose DSADiff (Data Structure-Aware Diffusion), a data structure-aware diffusion model that transforms the key factors identified by LAS into a unified generation difficulty representation and uses it to jointly drive feature modeling and training strategy optimization. This enables the model to adapt its representation and generation capability to samples with different data structures and difficulty levels, thereby mitigating learning bias across samples. Experiments on three CSP benchmark datasets, including Carbon-24, demonstrate that DSADiff consistently improves both Match Rate and RMSE, with a maximum Match Rate improvement of 9%, surpassing existing state-of-the-art methods and achieving new state-of-the-art performance. This work deepens the understanding of diffusion-based CSP models from the perspective of generation bias and provides a new direction for designing less biased generative models for complex materials spaces.
32. Every Measurement, Every Direction, All at Once: Multimodal Flow Matching for Molecules and Spectra
每一次测量、每一个方向、全部同时:面向分子与光谱的多模态流匹配
AI 总结:针对科学生成建模的数据限制问题,提出多模态流匹配模型MOSAIC,可实现分子与光谱间任意模态子集的生成,在QM9S数据集上的结构与光谱生成性能均优于基线模型。
链接:https://arxiv.org/abs/2611.22098
机构:Lila Sciences(莱拉科学公司)
作者:Thorben Prein, Elton Pan, Rafael Gómez-Bombarelli, Santiago Miret
英文摘要:Generative modeling in science is often data-limited, where experiments are expensive. However, each experiment may yield several coupled measurements of the same physical entity. Standard generative models often train only on a single direction, discarding supervision from the rest. We argue that each experiment contains more training signal than this asymmetry allows the model to exploit: random masking enables measurements to serve as either conditioning inputs or prediction targets, yielding multiple supervised denoising tasks from a single experiment. We investigate this premise with MOSAIC, a multimodal flow matching transformer over molecular 3D structure and UV, IR, and Raman spectra. MOSAIC is a single model that is capable of any-to-any generation: generating any subset of modalities from any disjoint subset of the others. On QM9S, MOSAIC achieves 88.8% molecular graph recovery, outperforming the strongest spectra-to-structure baseline by 23.2% in structure generation, while simultaneously outperforming a structure-to-spectra specialist model in spectral generation. Ablations show that the any-to-any objective improves structure generation over structure-only training across model sizes. Furthermore, MOSAIC supports arbitrary subsets of available measurements and enables molecular identification by matching generated spectra to experimental references.
33. Physically Admissible Generative Flow Matching for Sparse Flow Reconstruction
适用于稀疏流场重构的物理可容许生成流匹配
AI 总结:该研究针对稀疏流场重构的不适定逆问题,提出Feasible Subspace Consistency框架,通过切空间匹配等技术保证物理约束,在两类流场验证中提升模式覆盖、降低误差并消除谱畸变。
链接:https://arxiv.org/abs/2611.22549
机构:National University of Defense Technology(国防科技大学)
作者:Yuhao Wang, Yuanxi Peng
英文摘要:Reconstructing full fluid velocity fields from sparse, localized observations is a severely ill-posed inverse problem: most of the spatial domain remains unobserved, discrete conservation laws must hold strictly, and predictive uncertainty must be reliable. Existing generative models typically enforce physical constraints only at the terminal output or via soft penalties, leaving intermediate generative trajectories unconstrained. We introduce Feasible Subspace Consistency (FSC), a framework that preserves physical admissibility as an invariant of continuous generative dynamics: tangent-space matching confines the entire ODE trajectory to the discrete conservation manifold, while mass-orthonormal coordinates preserve the kinetic-energy metric. Sparse reconstruction then reduces to amortized posterior transport in the latent space of a frozen FSC prior, ensuring every posterior sample inherits incompressibility and boundary conditions without per-sample projection loops. We validate FSC on a periodic Kolmogorov flow, where exact spectral projection isolates path tangency and metric isometry, and a wall-bounded cylinder wake with immersed obstacles, realized via matrix-free iterative projection. FSC prevents the mode-coverage collapse induced by terminal projection (increasing mode coverage from $0.000$ to $0.790$) while eliminating high-wavenumber spectral distortion. In sparse reconstruction, amortized FSC reduces global relative error by $50\%$ and $46\%$, and sensor residuals by $63\%$ and $55\%$, relative to the unconditioned prior, while delivering calibrated, spatially heterogeneous uncertainty.
34. Path-Aligned Constraint Equivariance
路径对齐约束等变性
AI 总结:该研究提出PACE框架,通过零空间参数化和约束感知噪声采样,在不修改损失与积分器的情况下,精确执行物理场超分辨率的线性约束,大幅降低残差并恢复频谱与多样性。
链接:https://arxiv.org/abs/2611.22873
机构:National University of Defense Technology(国防科技大学)
作者:Yuhao Wang, Yuanxi Peng
英文摘要: Learned super-resolution of physical fields must satisfy structural constraints---such as divergence-free flow ($\nabla\cdot u=0$), curl-free fields ($\nabla\times u=0$), and pointwise sum-to-one normalization ($1^\top p=1$)---that standard training objectives do not enforce. Soft penalties trade fidelity against satisfaction via a scalar hyperparameter $\lambda$ but cannot drive the residual to zero; hard-constrained regression eliminates the residual by construction yet converges to the conditional mean, losing small-scale spectral structure. We observe that in Conditional Flow Matching with linear paths, a linear constraint $Ax=c$ is preserved along the entire transport trajectory whenever the velocity field is compatible ($Av_\theta=0$) and the initial noise is consistent ($Ax_0=c$); along the trajectory, the latter two conditions are also necessary. We term this the path-equivariance principle and instantiate it in PACE (Path-Aligned Constraint Equivariance), a framework that enforces these conditions through a null-space parameterization of the velocity field and a constraint-respecting noise sampler, with no modification to the loss function or the ODE integrator. Across three constraint families, PACE reduces the mean constraint residual by one to five orders of magnitude relative to unconstrained baselines, matching the residual level of the data-generation pipeline itself. The proposed framework requires only a closed-form parameterization onto the null space of $A$; within this scope, it offers exact constraint enforcement while recovering the spectral content and predictive diversity that conditional-mean regression systematically loses, all under a single unified recipe.
35. High-Dimensional Black-Box Optimization via Guidance Sampling of Diffusion Models
基于扩散模型引导采样的高维黑盒优化
AI 总结:提出BReaD框架,用GP作代理、UCB作引导,在低数据场景中有效优化高维黑盒目标,解决现有方法需标签或梯度的局限。
链接:https://arxiv.org/abs/2611.22965
机构:Robotics Institute, Carnegie Mellon University(卡内基梅隆大学机器人研究所); Princeton University(普林斯顿大学); Data and Systems Engineering, University of Hong Kong(香港大学数据与系统工程系); Princeton Plasma Physics Laboratory(普林斯顿等离子体物理实验室)
作者:Xiaoyan Hu, Rohit Sonker, Jalal Butt, Jiayu Chen, Egemen Kolemen, Jeff Schneider
英文摘要:Many real-world optimization problems involve optimizing black-box objectives for which only a limited number of evaluations are available. Bayesian optimization (BO) is a powerful framework for such problems, but its effectiveness often deteriorates in high-dimensional spaces. Recent work leverages diffusion models for black-box optimization, often through pre-training a reward-conditioned score network to steer generation towards high-reward samples. Alternatively, gradient-based guidance methods assume access to a differentiable reward function from which guidance gradients are queried. However, these approaches either require large amounts of labeled data for offline training or access to exact reward gradients, limiting their applicability in black-box settings. We propose BReaD (Black-box Reward-guided Sampling of Diffusion Model), a guidance sampling framework for optimizing black-box rewards without assuming access to a labeled dataset or reward gradients. The proposed method uses a Gaussian process (GP) as a differentiable surrogate for the unknown objective and actively queries labels for the generated samples. We provide a theoretical analysis of the total-variation distance between the generated distribution and the target distribution. Using Upper Confidence Bound (UCB) as guidance signals, our method balances exploration and exploitation during generation and enables the discovery of high-reward regions in the data space. Numerical experiments on scientific discovery and image generation tasks demonstrate that our framework can effectively optimize black-box rewards in low-data scenarios.
36. A Neural Envelope Method for Rapid Gradient-based Optimum Takeoff Trajectory Design
用于快速基于梯度的最优起飞轨迹设计的神经包络方法
AI 总结:该研究将physicsGAN与神经包络结合,将带约束非凸起飞轨迹优化转为无约束凸优化,在空客A3 Vahana eVTOL的优化中实现高精度与超快速收敛,大幅提升优化效率。
链接:https://arxiv.org/abs/2611.23130
机构:Missouri University of Science and Technology(密苏里科技大学)
作者:Samuel Sisk, Xiaosong Du
英文摘要:Urban air mobility arises to mitigate ground traffic congestion. Multidisciplinary analysis and optimization (MDAO) has been introduced to reduce the energy consumption, thereby expanding aircraft performance. Surrogate-based MDAO lowers the cost of optimization, but multimodality disrupts global optimum search for both simulation-based and surrogate-based optimizations. Convex envelopes convert multimodal spaces into unimodal spaces, permitting efficient convex optimization; however, convexifying a design space may produce infeasible "optimal" designs due to complex design constraints. To address these challenges, this work proposes the coupling of a physics-constrained generative adversarial networks (physicsGAN) and neural envelope to transform original, constrained, non-convex optimizations into unconstrained, convex optimizations. Specifically, the physicsGAN discovers a feasible design space where all design candidates inherently satisfy all constraints, by penalizing surrogate-predicted constraint violations during training. Moreover, the neural envelope constructs a neural network-based convex envelope for the transformed objective function within the physicsGAN-discovered feasible space. Note that the neural envelope only needs to deal with objective function since constraints are inherently satisfied in the feasible space. The neural envelope is demonstrated on takeoff trajectory optimizations of the Airbus A3 Vahana eVTOL aircraft. Results show that neural envelopes achieve a 99.91% mean accuracy within a flight-condition space compared with reference convex envelopes. Neural-envelope-enabled optimizations achieve 99.84% accuracy on the global optimum design, costing only 0.04 seconds on average. The neural envelope accelerates the trajectory optimization by 53 and 162 times on average compared to the physicsGAN-enabled gradient-based and global optimizations, respectively.
37. Correcting Flows with Marginal Matching
通过边际匹配校正流
AI 总结:本研究提出基于训练的流匹配预测-校正框架,通过边际匹配校正样本,在CIFAR-10和ImageNet-64上提升样本质量,且计算开销极小,解决流匹配模型效率与保真度平衡问题。
链接:https://arxiv.org/abs/2611.23134
机构:Technion – Israel Institute of Technology(以色列理工学院)
作者:Meital Bojan, Nadav Bojan Sellam, Ev Zisselman, Aviv Tamar
英文摘要: Flow matching models, ODE-based generative models, generate samples by gradually morphing a simple source distribution into a target distribution. However, these models still face challenges in optimally balancing computational efficiency with generation fidelity. While predictor-corrector schemes have been proven to help address this issue in diffusion models effectively, their application to flow matching models remains under-explored. In this work we present a novel training-based predictor-corrector framework for flow matching, which exploits the fact that the ODE can be easily reversed. Our key idea, termed Marginal Matching, is to correct samples at a particular time step in the forward ODE towards a matching marginal distribution of the backward ODE. This scheme proves to reduce a bound on the distance between the data distribution and the learned distribution, motivating two different implementations for the correction function. We show that our proposed method improves sample quality on CIFAR-10 and ImageNet-64, with minimal overhead in computation time.
5. 优化、泛化与理论分析 | 5 篇
38. When Does Discrete Ricci Curvature Help? An Edge-Level Analysis of Curvature-Enhanced GCNs for PPI Prediction
离散Ricci曲率何时有用?面向PPI预测的曲率增强型GCN的边级分析
AI 总结:本研究在四个PPI数据集上对比CGCN与标准GCN,发现CGCN增益集中在低度边、低曲率边性能下降,提出曲率加权增强簇间分离的新解释,并给出优先选择CGCN的实用指导。
链接:https://arxiv.org/abs/2611.21849
机构:ETH Zurich(苏黎世联邦理工学院); Max Planck Institute for Human Cognitive and Brain Sciences(马克斯·普朗克人类认知与脑科学研究所)
作者:Clara Joswig, Elham Gholamzadeh, Ghazal Mostofi, Nico Scherf
英文摘要:Curvature-enhanced graph convolutional networks (CGCN) incorporate Ollivier-Ricci curvature (ORC), grounded in optimal transport theory, into message passing, motivated by the claim that curvature-weighted edges alleviate "over-squashing" at structural bottlenecks. But does CGCN's improvement actually come from mitigating over-squashing? We investigate this on four real-world protein-protein interaction (PPI) datasets, comparing CGCN against a standard GCN at the level of individual edges rather than aggregate metrics alone. While CGCN improves aggregate AUC, AUPR, and F1 across all four datasets, this gain is concentrated almost entirely on low-degree edges, while performance on low-curvature (bottleneck) edges systematically degrades in three of four datasets, the opposite of what the over-squashing motivation predicts. Examining the learned edge-weight function directly, together with clustering-validity metrics, edge-distance ratios, and a layer-depth ablation across four depths, we find evidence for an alternative account: curvature-weighting sharpens the separation between clusters, which improves intra-cluster predictions at the cost of inter-cluster (bottleneck) ones. We test this account against topologically defined communities rather than the classification labels used elsewhere. Additionally, we offer practical guidance on when to prefer CGCN over standard GCN, depending on whether intra- or inter-cluster predictions matter most.
39. Weight-space programming for tractor guidance controller: large language models natively write neural network weights that steer a real tractor
拖拉机导航控制器的权重空间编程:大语言模型原生编写操控真实拖拉机的神经网络权重
AI 总结:本研究提出权重空间工程,用GPT-6 Astra等大语言模型直接编写拖拉机导航控制器的神经网络权重,在模拟与真实拖拉机测试中,其控制器无需干预即可导航,误差与纯追踪算法相当。
链接:https://arxiv.org/abs/2611.21958
机构:Technical University of Munich (TUM)(慕尼黑工业大学); Munich Institute of Robotics and Machine Intelligence (MIRMI), Technical University of Munich (TUM)(慕尼黑工业大学慕尼黑机器人与机器智能研究所); Aalto University(阿尔托大学)
作者:Timo Oksanen
英文摘要:Weight-space programming is a process where a large language model reads and writes the weights of a neural network directly, guided by a natural language specification, without an optimiser or gradient descent. This study applies it to automatic tractor guidance, where a 253-parameter neural network steers along paths of lines and arcs and must be adapted when the machine changes, such as a slower steering actuator. GPT-6 Astra and Claude Fable 5.1 (also 5.2) edited the network for our research tractor using a small budget of simulated evaluations to repair a network trained for another tractor and write a controller into a network with zero weights. We compared them with an evolution strategy, gradient descent and pure pursuit in simulation and on the real tractor. Both models read the network and analysed the closed loop with control theory correctly, and Fable's repair without simulation succeeded only occasionally. With simulated feedback, both models repaired the trained network in single sessions and wrote a working controller from zero weights, whereas from zero the evolution strategy found none. The controllers written from zero were sparse and in a classical cascaded structure; one model edited the trained network while the other replaced it. Gradient descent reached higher accuracy with far more simulation, and all language model controllers steered the real tractor without intervention, with errors of the same order as pure pursuit. Language models can program the weights of a neural controller directly, much as control engineers design one. We call this Weight-Space Engineering.
40. Large Language Models as Nonlinear Dynamical Systems: The Three Clocks of Depth, Training, and Generation
大型语言模型作为非线性动力学系统:深度、训练与生成的三个时钟
AI 总结:该研究将非线性动力学与语言模型领域关联,构建三时钟词典与降阶模型,在GPT-2等模型上验证,提出12个待解决的开放问题。
链接:https://arxiv.org/abs/2611.21983
机构:Leo AI, Inc.(Leo AI公司)
作者:Maor Farid
英文摘要: Language models are dynamical systems. A transformer is iterated over its layers, over the steps of its training and over the tokens it generates, and on each of these three clocks it shows behavior that nonlinear dynamics has studied for a century: fixed points, bifurcations, synchronization, chaos and noise-driven escape from potential wells. The two communities rarely meet, so this behavior is often rediscovered under new names. We build a bridge between them. To describe, we condense about 200 papers into the Three-Clock Dictionary, which maps fifteen concepts of nonlinear dynamics onto each clock and marks every correspondence as proved, measured or conjectured, and we show each concept in a mechanical system and in a language model side by side. To explain, we derive reduced-order models: the softmax is a Boltzmann distribution, gradient descent an Euler step on a nonlinear spring, a loss spike a ball thrown out of a stiff valley, attention on the circle a Kuramoto model, and a repetition loop a well that deepens while it is occupied. We then test these explanations on open language models, on a laptop CPU. Escape from repetition loops in GPT-2 and Pythia follows the Arrhenius law with the logit gap as the barrier, and the barrier deepens and then saturates, so the temperature needed to leave a loop falls like $1/\ln N$ with the generation length $N$ and rises with model size. Coherent text lives in a narrow window of temperatures, which a small reasoning model shares at its recommended temperature. Training makes each Pythia layer more expansive, and a small transformer trains at a multi-mode edge of stability. The third step, solving problems of language models, is still ahead: controllers derived from the escape analysis remove loops but do not yet beat the best fixed temperature. We state twelve open problems where nonlinear dynamics could get there.
41. Learning Optimal Transport Plans Via Autoregressive Token Regression
通过自回归 token 回归学习最优传输计划
AI 总结:本文提出基于 Transformer 的 ToR-OT 模型,将最优传输重构为序列生成任务,实现零样本泛化、稀疏解码与高效推理,为理解自回归模型学习约束优化结构提供了方向。
链接:https://arxiv.org/abs/2611.22218
机构:University of Wisconsin–Madison(威斯康星大学麦迪逊分校)
作者:Ivan Jaen Marquez, Sotirios Panagiotis Chytas, Vikas Singh
英文摘要:Optimal Transport (OT) provides a principled framework for finding the minimum cost coupling between probability distributions. Classical solvers remain the gold standard for this task. At the same time, OT is an ideal testbed for studying whether learning-based amortized models can internalize the structure of OT type problems. Here, ground-truth solutions are available and marginal constraints are explicitly defined. Also, exact discrete solutions have well-characterized sparsity. We describe ToR-OT, a transformer-based model that learns to predict discrete OT plans through autoregressive token regression. We reframe OT into a structured sequence generation task. ToR-OT decodes sparse transport plans as sequences instead of solving an optimization task at test time. We focus on three properties: (a) Zero-shot generalization: pretraining on a large synthetic corpus of OT instances allows a single model to generalize to unseen distribution pairs and heterogeneous cost metrics without retraining, (b) Sparse primal decoding: by exploiting the inherent sparsity in exact discrete OT solutions, the decoder predicts only nonzero entries (avoids dense coupling outputs), (c) Amortized inference: on low-dimensional discrete problems, ToR-OT gives accurate approximate plans and compares favorably with per-instance neural OT baselines. Overall, ToR-OT is not yet a replacement for classical solvers, but a step toward understanding how autoregressive transformers can learn structure of constrained optimization problems like OT.
42. A Transformer-Derived Iterative Preconditioner
一种源自Transformer的迭代预条件子
AI 总结:该研究从求解线性系统的Transformer中发现迭代预条件原语,开发出无需学习组件的RAPID算法,经实验验证其可改善矩阵条件数、减少共轭梯度迭代,且有收敛性理论保证。
链接:https://arxiv.org/abs/2611.22941
机构:Boston University(波士顿大学)
作者:Patrick Lutz, Themistoklis Haris, Aditya Gangrade, Venkatesh Saligrama
英文摘要:Transformers trained on in-context prediction problems can learn known numerical algorithms, e.g., gradient descent. We go further, showing that their internal computations can reveal new algorithms with provable guarantees. A softmax-attention transformer trained to solve linear systems $Xw=y$ implicitly improves matrix conditioning. Tracing $XX^\top$ reveals an iterative preconditioning primitive, which we develop into RAPID, a standalone algorithm requiring no learned components. RAPID is an anytime preconditioner for symmetric positive definite (SPD) matrices, using randomized mixing and sparse local corrections to improve conditioning. Proof-of-concept experiments on synthetic and real-world problems consistently improve conditioning and reduce conjugate gradient iterations needed to reach a target accuracy. For a Haar-mixing idealization, we prove convergence to any fixed target condition number regardless of spectral distribution, with only logarithmic dependence on initial conditioning. Together, these results show how a transformer can discover a broadly applicable numerical algorithm.
6. 高效学习、压缩与部署 | 2 篇
43. SilverSpoon-KD: A General-Purpose Toolkit for Knowledge Distillation
SilverSpoon-KD:一种通用的知识蒸馏工具包
AI 总结:该研究提出开源PyTorch工具包SilverSpoon-KD,统一多种知识蒸馏范式,实现通用分块蒸馏,经多类模型与任务验证,其性能优于或相当于现有工具包。
链接:https://arxiv.org/abs/2611.22842
机构:KTH Royal Institute of Technology(瑞典皇家理工学院)
作者:Xaver R. Davey, David Broman
英文摘要:Knowledge distillation transfers deep learning representations from a teacher model to a student model, typically reducing model size or inference cost. However, knowledge distillation tooling has not kept pace with the breadth of research methods: existing toolkits typically support only a limited subset of techniques and use cases. We present SilverSpoon-KD, an open-source PyTorch toolkit that unifies different distillation paradigms and architectural relationships across workloads and applications behind a single API. Moreover, SilverSpoon-KD offers the first general-purpose blockwise knowledge distillation implementation with unique scalability features. We validate its versatility across vision, masked language model, and causal language models on tasks spanning homogeneous, heterogeneous, and quantization distillation across paradigms. We also evaluate SilverSpoon-KD by comparing its performance, including throughput and GPU memory usage, with existing toolkits. Despite extensive support for convenience, metric calculation, and tracking, we have found SilverSpoon-KD's performance to be on par with or better than existing toolkits.
44. Ambient LLMs: Data-Efficient Supervised Fine-tuning via Temperature Matching
环境大语言模型:通过温度匹配实现数据高效的监督微调
AI 总结:该研究针对自回归语言模型微调的过拟合与灾难性遗忘问题,提出Ambient SFT框架,通过温度匹配利用预训练模型的不完美预测扩充微调集,在三类任务上较标准SFT获最高21%平均相对提升。
链接:https://arxiv.org/abs/2611.23229
机构:MIT(麻省理工学院); Arizona State University(亚利桑那州立大学); University of Texas at Austin(德克萨斯大学奥斯汀分校)
作者:Giannis Daras, Marios Papachristou, Adam Klivans, Antonio Torralba, Constantinos Daskalakis
英文摘要:Collecting large-scale, diverse, and high-quality post-training datasets for autoregressive language models is challenging. Consequently, Supervised Fine-Tuning (SFT) is typically applied to datasets significantly smaller than those used for pre-training, which leads to overfitting and the catastrophic forgetting of the tails of the data distribution. To prevent this collapse, we introduce Ambient SFT, a principled framework for leveraging imperfect predictions to augment the fine-tuning set. In our method, a "frozen" version of the pre-trained model serves as a noisy teacher that the fine-tuned model must match at a coarse level. In particular, we require that the high-temperature versions of the conditional distributions produced by the post-training model should match the high-temperature versions of the pre-training ones. The temperature hyperparameter controls the level of trust in the pre-trained model, with the limit $T\to \infty$ reverting back to regular SFT and $T=1$ to knowledge distillation. We demonstrate that our approach outperforms standard SFT across domain adaptation, instruction-tuning, and reasoning, leading for up to $21\%$ average relative improvement to standard SFT.
7. 联邦学习、隐私与安全 | 2 篇
45. Who Pays for Byzantine Robustness? Validity, Data-Share Dictatorships, and Fairness When Only a Few Hospitals Collaborate
谁为拜占庭鲁棒性买单?仅少数医院协作时的有效性、数据份额独裁与公平性
AI 总结:该研究针对仅3-6家医院协作的临床联邦学习场景,审计8种聚合规则,发现样本量加权会引发数据份额独裁,最大医院攻击时不同规则的性能下降原因存在差异,未加权规则会惩罚异常参数更新的机构。
链接:https://arxiv.org/abs/2611.22904
机构:Sapienza University of Rome(罗马第一大学)
作者:Adnan Malkani
英文摘要:Clinical federated learning (FL) collaborations typically connect only three to six hospitals, yet Byzantine-robust aggregation rules are designed and benchmarked on synthetic partitions across dozens of equal-sized clients. We conduct an empirical audit of eight standard aggregation rules: FedAvg, coordinate-wise median, trimmed mean, Krum, Multi-Krum, Bulyan, geometric median (GeoMed), and Centered Clipping (CC). We evaluate both unweighted and sample-size-weighted variants across three cross-silo medical datasets from the FLamby benchmark: Fed-Heart-Disease ($n = 4$ hospitals), Fed-TCGA-BRCA ($n = 6$ regions), and Fed-ISIC2019 ($n = 6$ centres). We execute 54,214 training runs across enumerated attacker coalitions ($f \in \{1, 2\}$) paired with exact-twin leave-out controls that decompose performance drops into honest data loss versus adversarial attack damage. We report three findings. First, client-count validity checks fail to capture breakdown in skewed medical consortia: while $n = 6, f = 1$ satisfies every count condition except Bulyan's, assigning sample-size weights turns median and GeoMed into a clean-model dictatorship whenever a single hospital holds more than half the training data, and causes weighted median, GeoMed, and CC to collapse when that dominant centre attacks. Second, exact-twin leave-out controls reveal that when the largest hospital attacks, the apparent failure of count-based robust rules is predominantly caused by the loss of honest training data (7.5 to 10.4 points on ISIC) rather than adversarial perturbation (2.3 to 3.8 points on ISIC). Conversely, weight-dependent rules suffer large true attack damage (16.1 to 25.1 points on ISIC). Third, unweighted robust rules consistently discount whichever institution generates the most atypical parameter updates, penalizing the smallest hospital on Heart but penalizing larger institutions on TCGA and ISIC.
46. Projected Penalty-Gradient Tracking for Communication-Efficient Byzantine-Resilient Federated Learning
面向通信高效且抗拜占庭攻击联邦学习的投影惩罚梯度跟踪方法
AI 总结:针对异构数据集上存在拜占庭攻击者的联邦学习,提出投影惩罚梯度跟踪框架及FRPG-CS、LAG-FRPG算法,在Min-Max攻击下准确率优于12种对比方法,同时平衡了抗攻击性与通信效率。
链接:https://arxiv.org/abs/2611.22929
作者:Yanjie Dong
英文摘要:We investigate communication-efficient federated learning over heterogeneous datasets in the presence of Byzantine adversaries capable of uploading arbitrary information. Penalty-based methods bound the influence of each worker through the bounded gradient of a penalty function; however, the bound is lost once the uploads are compressed, since a Byzantine worker can accumulate an unbounded state at the server. To retain the Byzantine resilience under compressed uploads, we propose a projected penalty-gradient tracking framework, where each honest worker uploads a compressed residual of its penalty gradient and the server projects each reconstructed state onto the known penalty-gradient ball. The framework accommodates any contractive compressor and yields two algorithms: fault-resilient proximal gradient with consistent sparsification (FRPG-CS), whose shared mask is protected by freshly drawn reserve rows, and lazily aggregated FRPG (LAG-FRPG), whose uploads are event-triggered. For smooth and strongly convex losses, we establish a finite-time bound on the expected objective gap, where the Byzantine error floor is independent of the compressor and the compression error decays faster than the optimization error. We also bound the expected number of lazy uploads and derive certificates that separate the finite-penalty model bias from the optimization error. Experiments on MNIST and Fashion-MNIST show that the proposed algorithms attain the highest accuracy among twelve methods under the Min-Max attack, whereas a compressed-momentum benchmark and a local-update benchmark remain preferable in attack-free accuracy and total traffic, respectively.
8. 鲁棒性、不确定性与可信学习 | 5 篇
47. A Temporal Algebraic Framework for Data Stream Evaluation Under Delayed Labels and Distribution Shifts
延迟标签与分布偏移下数据流评估的时态代数框架
AI 总结:该研究针对数据流的延迟标签与分布偏移问题,扩展代数框架提出时态加权方法,分离先验偏移与概念漂移,经实验验证其检测器在标签偏移下误报率低,代码已开源。
链接:https://arxiv.org/abs/2611.21914
机构:Universidade Federal do ABC (UFABC)(巴西联邦ABC大学); Universidade Federal de Uberlândia (UFU)(Uberlândia联邦大学)
作者:Ronaldo C. Prati, Elaine Ribeiro de Faria
英文摘要:In data streams, labels often arrive late or only for some instances, class prevalence drifts, and rare classes vanish from short windows. Standard evaluation then confuses a degraded decision boundary with a change in prevalence, and is biased when labels are delayed or selected by the model. We extend an algebraic framework that obtains pooled (micro) and per-class (macro) measures from one confusion matrix to streams, weighting the counts in time by exponential fading or a sliding window; the update costs O(m^2) per step, whatever the stream length, and each pooled measure stays a weighted combination of the per-class ones. Delayed and selective feedback enter as an observation mask: until its label arrives, an instance is counted with the model's calibrated class probabilities, and the bias of this interim estimate is driven by their calibration error. Audited conventions report no value for a class without verified instances. We separate prior shift from concept drift by monitoring prior-free coordinates (recall and specificity) and attribute a change in a measure to the two sources with a first-order Jacobian decomposition. On synthetic streams and four real datasets, a detector on these coordinates with a memory per class raises no excess false alarms under label shift, unlike the Jacobian alarm and error-rate detectors. The correction for delayed labels recovers recall in some real settings and is only partial for precision, specificity and F1. Twenty-five statements are checked in Lean 4, and the code is in the aggmetrics package.
48. How Much Distribution Shift Can a Clustering Withstand? Exact Split and Merge Certificates for Structured Transports
聚类能承受多大的分布偏移?结构化传输的精确拆分与合并证书
AI 总结:该研究针对聚类的分布偏移问题,基于确定性退火的局部稳定性准则,通过半定规划得到结构化偏移下的精确拆分与合并半径,构建了聚类的鲁棒性证书。
链接:https://arxiv.org/abs/2611.22000
机构:University of Illinois Urbana-Champaign(伊利诺伊大学厄巴纳-香槟分校); University of California, Los Angeles(加利福尼亚大学洛杉矶分校)
作者:Alisina Bayati, Srinivasa Salapaka, Shahriar Talebi
英文摘要:In clustering, the distribution of the population being clustered is often uncertain or changes over time, so the true distribution may shift away from the nominal distribution used to form the clustering. Once the clustering is committed and each member is assigned to a cluster, such shifts may change the distribution of each cluster. They can make a cluster too spread out for one representative at a given resolution, motivating a split, or bring distinct clusters close enough to share a representative at that resolution, motivating a merge. We make these split and merge events precise using a local stability criterion built on the phase-transition analysis in deterministic annealing. This notion of stability enables us to quantify how robust a clustering is against such events. Furthermore, motivated by applications in which uncertainty and population change are structured, we study shifts whose displacements are restricted to a prescribed subspace. The smallest structured shifts that reach the respective thresholds define the split and merge radii, for which we obtain exact semidefinite programming (SDP) reformulations. The split SDP has only two scalar decision variables. These convex reformulations enable us to combine the radii into a robustness certificate for the full clustering that gives the size of the smallest structured shift that reaches any monitored split or merge threshold.
49. Lost in the Funnel: Selection-Induced Miscalibration in Ranking and Decision Systems
迷失在漏斗中:排序与决策系统中由选择诱导的校准偏差
AI 总结:本文针对排序与决策系统中分数在漏斗流程使用时的校准偏差问题,提出漏斗感知重校准(FAR)方法,经线上A/B测试可显著降低预期校准误差并提升拍卖效率。
链接:https://arxiv.org/abs/2611.22595
机构:Microsoft Ads(微软广告)
作者:Aditya Somani, Rahul Ragesh, Vikas Joshi, Amit Singh
英文摘要:Modern ranking and decision systems combine components with divergent objectives, and scores calibrated at the model boundary are used as calibrated probabilities throughout the funnel, across filtering, reranking, business-rule enforcement, and auctions. Yet calibration is typically validated only where the score is produced, so its breakdown where the score is used is routinely overlooked. We ask when calibration survives this selection, how inherited error differs from selection effects, and what can fail for a perfectly calibrated score at a component boundary. We study these questions through a boundary-wise diagnostic framework of residual identities. We target rubric-judged relevance rather than presentation-dependent metrics such as CTR, so calibration is measurable at every funnel layer of interest. We first present an industry-wide case study that applies our diagnosis to settings where calibration deteriorates at decision time, tracing the degradation to selection rather than model error. This diagnosis motivates Funnel-Aware Recalibration (FAR): post-hoc, selection-policy-aware recalibration at the boundary where scores drive decisions. In a live two-region production A/B test, FAR cuts decision-time expected calibration error (ECE) by 60-80% and improves auction efficiency. We further adapt two public datasets to simulate industry settings, providing additional evidence for the mechanism. We plan to release the adapted datasets and their generation pipeline to support further research. Finally, we distill these findings into practical operating principles for calibrating multi-stage scoring systems.
50. The Price of Hard Coverage Constraints in Online Conformal Prediction
在线共形预测中硬覆盖约束的代价
AI 总结:本文研究在线共形预测中硬覆盖约束的效率代价,确定了不同场景下的最小累计超额预测集大小,证明利用过去得分可达到对应速率,且完美统计知识无法免除序列可靠性的代价。
链接:https://arxiv.org/abs/2611.22646
机构:George Mason University(乔治梅森大学)
作者:Mingrui Liu
英文摘要:Conformal prediction provides distribution-free uncertainty quantification, but in sequential deployment, standard coverage guarantees can still allow substantial fluctuations in the number of miscoverages over time. We study an online setting in which prediction sets are produced one round at a time and the cumulative number of miscoverages must stay within a prescribed budget throughout this http URL a hard coverage guarantee can be enforced by enlarging prediction sets, but how much additional prediction-set size is fundamentally necessary remains unclear. We characterize this unavoidable efficiency cost over $T$ prediction rounds. Remarkably, the cost persists even with i.i.d.\ data, a known score distribution, and perfect knowledge of the population-optimal prediction set. Under suitable regularity conditions relating coverage and prediction-set size, we determine the minimum cumulative excess prediction-set size required by any online method satisfying the hard coverage constraint. This minimum cost is $\Theta(\log T)$ for expected cost at a fixed time known in advance, $\Theta(T^{1/3})$ for the largest expected cost over all fixed times up to $T$, and $\Theta(\sqrt{T})$ for the expected worst realized cost when the evaluation time is chosen in hindsight. We further show that these rates can be attained using past scores, without knowing the score distribution. Thus, even perfect statistical knowledge does not make sequential reliability free.
51. Get Hint-Augmented GRPO Done Right Through Activation Steering
通过激活引导正确实现提示增强型GRPO
AI 总结:本研究针对现有提示增强型GRPO存在的训练-推理上下文不匹配问题,提出基于激活引导的ASH方法,通过训练引导向量在推理时重现提示行为,在多个数学基准上显著提升了LLM的pass@k性能。
链接:https://arxiv.org/abs/2611.23124
机构:University of Notre Dame(圣母大学)
作者:Khiem Le, Ting Hua, Nitesh V. Chawla
英文摘要:GRPO has become the dominant reinforcement learning method for eliciting reasoning capability in LLMs using verifiable rewards. However, it suffers from a fundamental failure mode: when a training question exceeds the base model's current capability, all generated responses fail and receive zero reward, yielding no gradient signal and stalling further capability expansion. An emerging class of GRPO extensions, termed Hint-Augmented GRPO (HA-GRPO), adds a "hint" to the context to reduce the difficulty of the question, thereby producing a usable gradient signal. Nevertheless, existing HA-GRPO methods share a key issue: no hint is available at inference time to append to the question, creating a training-inference context mismatch that leads to inconsistent performance gains. In this work, we reinterpret the function of hinting -- not as reducing question difficulty, but as steering model behavior -- and propose ASH (Activation Steering for Hinting), a novel HA-GRPO method built on this view. During training, the model is presented with both hinted and plain questions, while a trainable steering vector is optimized to widen the margins between the model's generation probabilities for responses to the hinted version relative to the plain version, thereby capturing the question-agnostic component of the behavioral shift induced by the hint. During inference, the trained steering vector is directly injected into the model's activation, reproducing the hint-conditioned behavior without requiring any hint in the context. Experiments with four LLMs (Qwen3-1.7B/4B-Base, Llama-3.2-1B/3B-Instruct) show that ASH consistently improves pass@$k$ on competition-level mathematics benchmarks (AMC, OlympiadBench, AIME24, AIME25) and out-of-distribution benchmarks (Minerva, GPQA-Diamond). Notably, it improves average pass@8 of Qwen3-1.7B-Base and Qwen3-4B-Base by \textbf{2.60}\% and \textbf{2.21}\%.
9. 图学习与结构化数据 | 1 篇
52. DLS: Dual-Logic Graph Neural Networks with Chaos-Fuzzy Residuals for Robust Link Prediction
DLS:带有混沌-模糊残差的双逻辑图神经网络,用于鲁棒链接预测
AI 总结:本文提出DLS(双逻辑SAGE)框架,通过混沌注入模块与模糊逻辑门控机制突破图神经网络过平滑瓶颈,在异质图链接预测任务中性能优于现有方法,同时保持同质图上的效率。
链接:https://arxiv.org/abs/2611.22607
机构:University of Sistan and Baluchestan(锡斯坦-俾路支斯坦大学); University of Tehran(德黑兰大学)
作者:Mohammad Mehdi Keikha, Abolfazl Nadi
英文摘要:Link prediction is a fundamental task in graph mining, critical for applications ranging from biological interaction discovery to recommender systems. While Message Passing Neural Networks (MPNNs) have achieved state-of-the-art results, they are fundamentally limited by the inadvertent suppression of high-frequency signals, as the recursive aggregation that stabilizes structural learning inherently leads to over-smoothing, rendering node representations indistinguishable in deep layers. This limitation is particularly severe in heterophilous graphs, where structural proximity does not imply feature similarity. In this paper, we propose DLS (Dual-Logic SAGE), a novel inductive framework designed to break the over-smoothing barrier via a "controlled instability" mechanism. Unlike conventional methods relying on static structural heuristics, DLS introduces a dynamic residual pipeline. First, a Chaotic Injection Module utilizes the Logistic Map to diversify the feature manifold, exploiting sensitivity to initial conditions to separate topologically similar but semantically distinct nodes. Second, to mitigate the aleatoric uncertainty introduced by chaotic dynamics, a Fuzzy-Logic Gating Mechanism applies soft-thresholding to filter noise and retain only high-confidence signals. Experimental results on diverse benchmarks demonstrate that DLS effectively balances feature discrimination with robustness, significantly outperforming state-of-the-art baselines in heterophilous settings while maintaining competitive efficiency on standard homophilous graphs.
10. 迁移、元学习与持续学习 | 5 篇
53. KVUse: Reusable-KV User Memory for Efficient LLM Recommenders
KVUse:用于高效LLM推荐系统的可复用KV用户内存
AI 总结:KVUse将用户画像存为可复用KV缓存,通过LoRA和定制调度器降低LLM推荐系统的预填充开销,在不损失排序质量的前提下提升了评分与画像更新路径的吞吐量。
链接:https://arxiv.org/abs/2611.21920
机构:Meta Recommendation Systems (MRS)(元推荐系统(MRS))
作者:Xin Guo, Qiang Zhang, Xiang Zhou, Arushi Rai, Siyu Lin, Yifan Wu, Xiang Shen, Lei Huang, Qunshu Zhang, Xiangjun Fan, Lizhu Zhang
英文摘要:Memory-augmented recommenders built on large language models (LLMs) maintain a natural-language profile of each user. These profiles are updated by a user profile update path that revises the profile as new interactions arrive and are consumed by downstream task paths like ranking. Because these user profiles are stored as text in the user memory, every ranking LLM call that consumes them must splice them in the prompt and calculate the KV cache of them during the prefill phase. Given these ranking LLM calls are prefill heavy (long prompt but short decoded output) and repeatedly called, these repeated prefill of user profile text dominates the serving. Prefix caching cannot remove it, because these user profiles are preceded by different text in different calls. We present KVUse, which stores the user profiles as reusable key-value (KV) caches instead of text. KVUse trains lightweight LoRA writers by self-distillation, so that the user profile update call produces the reusable KV of the user profile while decoding it, and the LLM can read these reusable KV at new positions and behind new context without recomputation. KVUse also has a customized KV scheduler that uploads the stored reusable KV of the next batch while the current batch is computed, hiding the KV transfer from host memory behind GPU computation. On four InstructRec datasets with Llama-3.1-8B-Instruct, KVUse shows no statistically significant difference in ranking quality from full prefill over continuous memory update rounds. Under saturated load, KVUse raises the throughput of the downstream scoring path by 4.0x-4.5x and of the profile update path by 1.4x-1.9x.
54. BQ-LoRA: Binary-Quantized Low-Rank Adapters as Implicit Regularizers for Parameter-Efficient Fine-Tuning
BQ-LoRA:作为参数高效微调隐式正则化器的二值量化低秩适配器
AI 总结:针对参数高效微调量化大模型的QLoRA方法,提出BQ-LoRA将适配器参数二值化并结合缩放直通估计器,理论上刻画其假设类性质,在Qwen2.5等模型上实现存储大幅缩减且性能优于QLoRA。
链接:https://arxiv.org/abs/2611.22334
机构:KAIST(韩国科学技术院)
作者:Juyoung Park, Heejun Lee
英文摘要:Parameter-efficient fine-tuning of quantized large language models is dominated by QLoRA, which combines 4-bit weight quantization with FP16 low-rank adapters. We ask whether the FP16 adapter precision is necessary. Replacing it with a single binary bit per adapter parameter, combined with a learned per-layer scale $s$ and a Scaled Straight-Through Estimator, reduces adapter storage by 13-16x. We introduce two variants -- NormScale, which binarizes under a norm-preserving scale with a clipped STE, and LearnThresh, which makes the binarization threshold learnable -- unified under a framework we call BQ-LoRA. Our main theoretical contribution is a tight characterization of the BQ-LoRA hypothesis class: it has bounded entrywise norm, a typical-set spectral norm bound that is $\Theta(r)$ times sharper than the worst case, a full-rank, near-isotropic update, and consequently a Rademacher complexity that yields an explicit $\tilde{O}(s_{\max}\alpha\sqrt{d_{out}/(rn)})$ generalization bound. FP16 LoRA has no comparable a priori norm or isotropy control. On Qwen2.5 (1.5B-72B), BQ-LoRA exceeds QLoRA's benchmark averages at every scale by 0.8-3.5 points; on LLaMA-2 (7B/70B) it remains competitive at the same storage reduction but does not uniformly lead the strongest FP16-adapter baselines (LoftQ at 7B, QLoRA at 70B). On Qwen2.5-7B the binary adapter preserves the 4-bit base WikiText-2 PPL within 0.06 (9.86 vs. 9.80) while QLoRA worsens it by 0.59. The reduction applies to inference-time adapter storage and serving; training memory matches QLoRA and step time rises by under 4%. We translate the storage saving into a deployment study, BQ-SGMV, by extending Punica's SGMV kernel with a binary inner-product primitive: the same A100-80GB holds 15.6x more adapters in HBM, and our research prototype eliminates cache misses under a skewed synthetic workload.
55. Variance Restoration over Covariance Rank: Memory-Budgeted Analytic Continual Learning
协方差秩上的方差恢复:内存预算下的解析式持续学习
AI 总结:该研究针对解析式持续学习的内存瓶颈,提出CoRe方法,通过保留关键统计量、压缩非对角交互并恢复方差,在削减90.7%内存的同时保持与全协方差读出相当的性能,且在多数内存配置下优于窄读出。
链接:https://arxiv.org/abs/2611.23040
机构:University of Wisconsin–Madison(威斯康星大学麦迪逊分校); University of Washington(华盛顿大学)
作者:Elena Voss, Pixel Nomand, Marcus Hale, Sofia Reyes
英文摘要:Analytic continual learning replaces iterative gradient updates with closed-form ridge regression over frozen pre-trained representations, summarizing all past sessions in two running sufficient statistics: the feature autocorrelation matrix $\Sigma_t$ and the class cross-correlation matrix $\Gamma_t$. Lifting backbone features into a wide random code of dimension $M \gg D$ substantially improves linear separability across tasks, but forces $\Sigma_t$ to grow quadratically as $\mathcal{O}(M^2)$. Under a fixed persistent memory envelope $\Bcal$, a learner faces a fundamental structural trade-off: store exact pairwise co-activations inside a narrower feature code, or expand to a wide code while compressing cross-coordinate interactions. We observe that the coordinate self-variances $v_t = \diag(\Sigma_t)$ and class statistics $\Gamma_t$ grow only linearly with $M$, whereas off-diagonal interactions dominate memory. We introduce \method (\textbf{Co}ordinate-\textbf{Re}stored \textbf{A}nalytic \textbf{C}ontinual \textbf{L}earning), which retains $(v_t, \Gamma_t)$ exactly, maintains off-diagonal co-activations through a streaming rank-$q$ factor $F_t$, restores exact diagonal variances via $\Sighat_t = F_t^\top F_t + \Diag(v_t - \diag(F_t^\top F_t))$, and solves the full $M\times C$ normal equations via the Woodbury identity. Across five class-incremental benchmarks on frozen ViT-B/16, \method matches dense full-covariance readouts within $0.09$ points while cutting persistent state by $90.7\%$ ($35.9$ vs.\ $385.3$ MiB). Under equal memory envelopes of $16$--$80$ MiB, prioritizing expansion width $M$ over covariance rank $q$ outperforms exact narrow readouts in $17$ of $18$ configurations by up to $+2.19$ points.
56. Learning Stratified Policies for Inference-Time Objectives
面向推理时目标的分层策略学习
AI 总结:该研究提出分层策略,通过共享语言模型主干的独立LoRA适配器组成策略群体,优化pass@k,在多推理任务上相比单一策略基线提升了pass@k,证明分配样本给互补策略比重复采样更有效。
链接:https://arxiv.org/abs/2611.23154
机构:Boston University(波士顿大学); Broad Institute of MIT and Harvard(麻省理工学院与哈佛大学博德研究所)
作者:Mridul Mahajan, Xuezhou Zhang, Aldo Pacchiano
英文摘要:Inference-time scaling improves the performance of large language models by sampling multiple responses and evaluating their collective utility. However, many existing approaches obtain these samples by repeatedly sampling from a single policy. In this work, we study if the sample budget at inference time can instead be allocated across a population of policies. We introduce stratified policies, where each of $k$ learned policies contributes one response to a $k$-sample inference-time objective. We realize this policy population using separate LoRA adapters on a shared language model backbone and optimize it for pass@$k$. We theoretically motivate the approach by analyzing the linear setting; a population of LoRAs can achieve zero best-of-$k$ approximation error in settings where every single LoRA has strictly positive prediction error. Empirically, we demonstrate that stratified policies consistently improve pass@$k$ over a single-policy baseline across multiple reasoning tasks. Overall, our results highlight how distributing samples across complementary policies can be a more effective use of sample budget than repeated i.i.d. sampling from a single policy at inference time.
57. Fine-Tuning Gemma 4 Edge Models for MATLAB Signal-Processing Code with SFT and QLoRA on a Laptop iGPU
在笔记本电脑iGPU上通过SFT和QLoRA微调Gemma 4边缘模型以适配MATLAB信号处理代码
AI 总结:该研究在笔记本iGPU上用SFT和QLoRA微调Gemma 4边缘模型,适配MATLAB信号处理代码,提升了DSP规范下的代码通过率,且适配器运行成本可控。
链接:https://arxiv.org/abs/2611.23205
作者:Mehmet Altan Toksöz
英文摘要:Small language models that run on a laptop are attractive for writing signal-processing code offline, but they make systematic errors on digital signal processing (DSP) conventions. We apply supervised fine-tuning (SFT) to the Gemma 4 edge models E2B and E4B with 66 checked examples covering 11 signal-processing MATLAB subtasks, using low-rank adaptation (LoRA) and quantized LoRA (QLoRA) on the integrated GPU (iGPU) of a laptop, and deploy the 8-bit integer (int8) models with the adapters through OpenVINO. On held-out variants scored by deterministic checkers, the pass rate rises from 18/66 to 37/66 (LoRA) and 37/66 (QLoRA) for E2B and from 32/66 to 47/66 for E4B (QLoRA). Reading all 159 failing programs shows that the adapters remove most errors of convention, such as the sign of a lag, units and array shapes, but few errors in building signals and noise. Applied at run time, the adapter costs 17-19% of the generation speed.
11. 数据集、基准与评测 | 5 篇
58. Solver to System: LLM Reasoning Across Representations and Execution in Scientific Simulation
从求解器到系统:科学模拟中跨表示与执行的大语言模型推理
AI 总结:本研究提出Solver2System数据集,评估LLM跨表示一致性及物理无效性检测能力,发现LLM在证据冲突时偏向特定表示,执行可改善有效性判断但存在因果诊断不足问题。
链接:https://arxiv.org/abs/2611.21964
机构:New York University(纽约大学)
作者:Eliza Berman, Rohan Subramanian, Shreya Jha, Lohit Jagarapu, Romain Lopez, Rajesh Ranganath
英文摘要:Simulators based on partial differential equations (PDEs) are widely used to model physical systems, but generating and interpreting PDE solver code with large language models (LLMs) remains challenging. Scientific simulation code is a useful testbed for LLM scientific reasoning because (1) one system can be represented by code, equations, natural language, and numerical trajectories and (2) code can execute successfully while implementing invalid physics. We introduce Solver2System, a dataset of PDE solvers with multiple representations, controlled corruptions, and lexical perturbations, to evaluate whether LLMs can recover the underlying physical system through (1) cross-representation consistency and (2) static and agentic detection of physical invalidity. Across eight open-weight models, LLMs systematically favor some representations when evidence conflicts, and lexical perturbations to code alter interpretation of other unchanged representations. Execution improves physical-validity judgments, but weaker agents can judge correctly without identifying the causal defect. Thus, LLMs may reach correct judgments despite inconsistent representation reconciliation or causal misdiagnosis.
59. VTT: Reading Vehicle Size and Powertrain Type from Motion Trajectory Across Unseen Car Models
VTT:从未见车型的运动轨迹中读取车辆尺寸和动力系统类型
AI 总结:提出VTT数据集与基准,结合辅助训练等方法,从未见车型运动轨迹中识别车辆尺寸和动力系统,提升了分类准确率,验证了物理特征与识别任务的关联。
链接:https://arxiv.org/abs/2611.22109
机构:University of South Florida(南佛罗里达大学); NVIDIA(英伟达公司); Amazon(亚马逊公司); University of Florida(佛罗里达大学)
作者: Yuhang Wang, Yuechen Guo, Jingxin Yang, Chuheng Wei, Suyang Chen, Abdulaziz Alhuraish, Yiyi Wang, Mingyang Ban, Jingran Sun, Hao Zhou
英文摘要:Vehicle size and powertrain information can enrich trajectory data for traffic analysis and simulation. We present VTT, a dataset and benchmark for identifying these attributes from motion and controller area network (CAN) signals across unseen car models. The dataset contains 9,062 drives, about 3,950 h, from 1,431 vehicles and 174 car models. It provides three footprint-based size classes, powertrain labels, and evaluation on unseen car models and unseen drivers. Using 60-s windows, we compare four sequence architectures, statistical features and physical features proposed by large language models. Training with auxiliary physical targets, data augmentation and task-specific weighting raises drive-level macro-F1 to 92.6-93.0% for battery-electric versus combustion vehicles and from 62.9% to 67.1% for size classification. On public logs from 4,061 additional drivers, the full-input model identifies powertrain correctly for 86 of 88 car models; size accuracy is 63.2%, close to the internal vehicle-level result of 63.5%. Physical-feature analysis links powertrain recognition to deceleration behaviour and size recognition to steering response. Errors near size-class boundaries and limited gains from longer observations highlight the importance of label resolution and training-fleet coverage for cross-model generalisation.
60. SOTA or Luck? The Winner's Curse in LLM Leaderboards
SOTA还是运气?大语言模型排行榜中的赢家诅咒
AI 总结:该研究发现LLM排行榜存在赢家诅咒的排名诱导偏差,提出N选m自助法校正偏差并构建RULER报告,可量化排名区间、划分模型层级,提升排行榜评估的可靠性。
链接:https://arxiv.org/abs/2611.22227
机构:Olin Business School, Washington University in St. Louis(华盛顿大学圣路易斯分校奥林商学院); W. P. Carey School of Business, Arizona State University(亚利桑那州立大学W.P.凯瑞商学院)
作者:Sikun Xu, Raphael Thomadsen, Dennis J. Zhang, Heng Zhang
英文摘要:Public LLM leaderboards guide model evaluation and adoption, yet the reported performance of top-ranked models can be systematically overstated. This leaderboard winner's curse arises because ranking noisy benchmark scores rewards sampling luck alongside genuine ability. The same selection effect can distort scores throughout the leaderboard, a phenomenon we call the rank-induced bias. We develop an $m$-out-of-$N$ bootstrap to correct for this bias. It assumes a fixed model pool with independent, identically distributed bounded tasks while allowing arbitrary within-task correlation among models. Under exact population tiers separated by positive gaps, we prove that the bootstrap reduces $\Theta(N^{-1/2})$ ranking bias to expected residual bias $o(N^{-1/2})$. Across 14 benchmarks from 5 benchmark families, the rank-1 adjustments range from approximately 0.09 to 3.86 percentage points. In our SWE-bench Verified snapshot, the tied leaders receive an adjustment of 0.90 percentage points. Such overstatement is large enough to make the apparent winner statistically fragile. Thus, we use the same bootstrap to further quantify the intervals of plausible rank positions for each model and group models into tiers. We establish sufficient conditions for consistent recovery of population tiers. On SWE-bench Verified, this procedure places 13 of 173 models in the leading tier. We combine the score adjustment and rank intervals in the Rank Uncertainty and Leaderboard Evaluation Report (RULER). We validate its finite-sample performance using sample splitting and calibrated simulations, and advocate publishing RULER alongside raw rankings so readers can assess both reported performance and the precision of the ordering. RULER also provides guidance for benchmark design that accounts for selection effects and ranking uncertainty.
61. Encoding, Decodability, and Use: A Five-Layer Framework for Benchmark Measurement Audits with Evidence from UAV Intent Recognition
编码、可解码性与使用:用于基准测量审计的五层框架——以无人机意图识别为例
AI 总结:该研究提出用于基准测量审计的五层框架,在AeroIntent上实例化后发现基准精度未反映模型对径向速度的读取,为无人机意图识别的基准评估提供了新方法。
链接:https://arxiv.org/abs/2611.22434
机构:Shandong Management University(山东管理学院); Tsinghua University(清华大学)
作者:Zhirong Su, Tian Zhang
英文摘要:Benchmark accuracy is routinely read as evidence that a model understands the construct the benchmark measures. We propose and instantiate on AeroIntent a five-layer framework that decomposes that reading into observation, representation, decoding, use, and behaviour. The framework separates encoding, decodability, and use as distinct levels. In the instantiation, the observation layer is unreliable on 7 of the 20 Drone programs and floors the direct visual correlate of radial velocity at a 0.26\% scale change. The representation layer exposes two task-designed RGB features with nominal signals. RAFT flow radial bias reaches $r=+0.257$ ($p=0.030$), and bounding-box optical-flow divergence reaches $r=+0.203$ ($p=0.029$). No multi-feature readout gives a fold-mean test correlation above the cluster minimum detectable effect (MDE), and the largest positive global out-of-sample correlation is $r=+0.080$. The decoding layer sets the MDE at $|r|=0.599$ at the multi-feature Bonferroni threshold ($\alpha=0.00714$, power 0.80), and the observed $r=+0.080$ has power 0.00. Two injection experiments establish the MDE's condition from both sides: aligned injections are detected at or above the simulated power, and orthogonal injections are not detected. The use layer finds accuracy consistent with program memory (program-identity 0.937/0.597/0.210) and multi-region static cues. The behaviour layer is consistent with accuracy not requiring a radial-velocity readout (appearance 0.6552 vs bounding-box trajectory 0.4080, cluster-level paired $\Delta=+0.2471$, $p=0.0132$).
62. The Kernel Reality Check: Benchmarking and Distilling Efficient Attention at Scale
核方法验证:大规模高效注意力机制的基准测试与蒸馏
AI 总结:本研究针对核化注意力缺乏系统评估的问题,通过基准测试发现FlashAttention效率优势及质量差距,提出LARA++可恢复近99%准确率但仍存效率瓶颈。
链接:https://arxiv.org/abs/2611.22947
机构:Concordia University(康考迪亚大学); Mila – Quebec AI Institute(米拉-魁北克人工智能研究所); Laval University(拉瓦尔大学); University of Tübingen(蒂宾根大学)
作者:Fırat Öncel, Cem Subakan, Mirco Ravanelli, Çağatay Yıldız
英文摘要: Kernelized attention methods, which aim to replace the $O(N^2)$ complexity of softmax attention with $O(N)$ alternatives, have been the focus of intense research in the last five years. Despite growing interest, systematic evaluations and fair comparisons across approaches and scales are still missing. We analyze kernelized attention in two halves and highlight prominent limitations on each. On efficiency, FlashAttention achieves lower memory consumption and faster wall-clock time than SOTA kernelized methods during training and in vision transformers across most sequence lengths (with inference-time memory in causal language models the only clear exception), contradicting the core assumption for pure kernelized attention used as a drop-in replacement for softmax. We also discover that reported efficiency gains often fail to reproduce against modern baselines. On quality, current kernelized methods leave gaps of 2-10 points relative to softmax under a unified distillation protocol, and model scala alone does not close these gaps. We show this gap is not intrinsic to kernelization: LARA++, a principled extension of LARA that approximates softmax via importance sampling with adaptive proposal distributions, recovers up to 99% of softmax accuracy. We establish that the gap is closable; however, the computational overhead remains substantial, leaving the efficiency wall intact. Code is available at this https URL
12. 机器学习应用 | 6 篇
63. Physics-informed neural networks for spacecraft trajectory optimization with exact boundary conditions
用于带精确边界条件的航天器轨道优化的物理信息神经网络
AI 总结:针对标准物理信息神经网络处理航天器轨道优化时边界条件约束导致的收敛差、精度低问题,提出一种先验精确施加边界条件的变换,将问题转化为单目标无约束优化,有效提升了收敛性与求解精度。
链接:https://arxiv.org/abs/2611.21933
机构:Brown University(布朗大学); Purdue University(普渡大学)
作者:Samuel Schönherr, Günther Waxenegger-Wilfing, Sascha Ranftl
英文摘要:We use neural networks to optimize trajectories by modeling the position as a function of time and minimizing the thrust with respect to the network parameters. This corresponds to an unsupervised training procedure in which physical laws are incorporated directly into the optimization objective. In standard physics-informed neural networks, initial and terminal states are enforced through additional weighted loss terms, resulting in a multi-objective optimization problem that is highly sensitive to the choice of loss weights and can lead to poor convergence and reduced accuracy. To address this challenge, we construct a simple transformation, exploiting the mathematical structure of trajectory optimization problems, to enforce boundary conditions exactly a-priori. This transformation can be interpreted as a reparameterization of the solution space that embeds the boundary conditions directly into the function. This converts the original constrained optimization problem into an unconstrained one over admissible trajectories. I.e. the transformation eliminates the boundary condition penalty terms and reduces the problem to a single-objective optimization. Crucially, trajectory optimization comes with a low-dimensional boundary, which allows particularly simple and stable constructions of distance functions compared to general domains in partial differential equations. Numerical experiments demonstrate that the proposed approach meaningfully improves convergence behavior and solution accuracy.
64. Higher-Order Interaction Reconstruction via Integral Physics-Informed Neural Networks
基于积分物理信息神经网络的高阶交互重构
AI 总结:针对复杂网络高阶交互重构的逆问题,提出积分物理信息神经网络(IntegralPINN)框架,通过积分形式抑制噪声与不稳定性,结合交互张量连续参数化实现准确鲁棒重构,性能优于现有方法。
链接:https://arxiv.org/abs/2611.21935
机构:University of North Carolina at Chapel Hill(北卡罗来纳大学教堂山分校); University of Louisville(路易斯维尔大学); Illinois Institute of Technology(伊利诺伊理工学院)
作者:Xin Mao, Zhen Xu, Shiyi Yang, Sukwon Yun, Yunmei Liu, Ren Wang, Tianlong Chen, Can Chen
英文摘要:Reconstructing higher-order interactions is a fundamental challenge in complex networked systems, with applications in ecological, social, and biological networks. This inverse problem is inherently difficult due to the combinatorial complexity of higher-order structures, the strong nonlinearity of the underlying dynamics, and the prevalence of noisy and sparse observations. In this article, we propose a novel integral physics-informed neural network (IntegralPINN) framework for higher-order interaction reconstruction. Instead of { penalizing derivative-based physics residuals,} we adopt an integral formulation over temporal intervals, which naturally suppresses high-frequency noise and mitigates derivative-induced instability. We further introduce a continuous parameterization of interaction tensors, enabling tractable identification of higher-order structures via gradient-based optimization. We provide theoretical analysis establishing a noise-robust error bound that is independent of derivative amplification when the integration window satisfies an excitation condition. Extensive experiments on ecological dynamics, social contagion, and neuronal synchronization systems demonstrate that IntegralPINN achieves accurate and robust reconstruction under noisy and sparse observations, outperforming state-of-the-art approaches.
65. Synthetic Crystal Data Generation for Diffusion Pretraining in Material Property Prediction
面向材料属性预测中扩散预训练的合成晶体数据生成
AI 总结:本研究将扩散预训练与合成晶体数据生成结合,引入含三种策略的流水线,用机器学习势替代DFT弛豫,在7项MatBench任务及mp_gap任务上,合成预训练表现优于随机初始化,数据规模扩展可提升性能。
链接:https://arxiv.org/abs/2611.22140
机构:Georgia Institute of Technology(佐治亚理工学院)
作者:Aamod Varma, Shuyi Jia, Victor Fung
英文摘要: Diffusion-based pretraining has emerged as a powerful strategy for learning transferable representations in graph neural networks for materials property prediction without requiring quantum-mechanical labels such as energies, forces, and stresses for every pretraining structure. However, the method retains a critical dependency: it requires structures residing near local minima of a potential energy surface, which have traditionally been obtained through costly DFT relaxations of known or physically synthesized compounds. This reliance constrains both the scale and the structural diversity of available pretraining data, leaving large regions of crystal structure space unexplored. In this work, we connect diffusion-based pretraining with synthetic structure generation to reduce this bottleneck. We introduce a synthetic data generation pipeline combining three complementary strategies: perturb-and-relax, generative sampling via MatterGen, and protostructure enumeration. Where structural relaxation is required, we use fast machine-learned interatomic potentials instead of performing a new DFT relaxation for each generated structure. We further compare a coordinate-denoising control with continuous-time variance-preserving (VP) and sub-VP diffusion objectives using a randomly initialized Orb-v2 backbone. The source comparison covers seven MatBench property-prediction tasks, and an eighth task, \texttt{mp\_gap}, is included in the protostructure scaling study. Every synthetic-pretraining condition numerically outperforms random initialization on all seven source-comparison tasks. Scaling the protostructure corpus from 20k to 165k under a fixed 400-epoch schedule numerically matches or exceeds MP-20k pretraining on most tasks; because this protocol increases both data diversity and optimization updates, the result reflects joint data-and-compute scaling.
66. Testing World Models on Actions They Have Never Seen: Held-Out Plans, Forecast Accuracy and Price Rules in Retail
在从未见过的动作上测试世界模型:零售中未见过的计划、预测准确性与价格规则
AI 总结:该研究针对零售场景等领域,测试了世界模型对未见过动作的预测可信度,发现预测准确性在不同模型间的指导作用有限,价格规则可提升模型性能,且测试适用于所有需处理历史外动作的世界模型领域。
链接:https://arxiv.org/abs/2611.22537
作者:Karthik Iyer
英文摘要:A world model is trusted to answer "what if": how its world responds to an action it has never seen. We give a test of that trust for any action-conditioned world model: actions held out of training by design and scored against a known answer; whether forecast accuracy predicts accuracy on those actions; and the same structure given to simpler models. We run it in retail, where the action is a price and promotion plan, across 2,853 fitted models. First, forecast accuracy is a fair guide between kinds of model (rank correlation with error on new plans 0.32 to 0.66) and a poor guide between versions of one world model: where merchants plan into high-demand weeks the link reverses, down to -0.69. Second, training the world model to learn the effect of actions (JEPA pretraining, a latent consistency check, adaptation at the plan date) leaves its error on new plans unchanged. Third, in generated worlds the recorded regular price and three price rules improve it. Given the same rules, a log-linear model is best when its curve can match the truth; in a world whose curve lies outside the rules, the world model with the rules is ahead of the log-linear model with the rules in 36 of 40 worlds, and does as well without its recurrent cell. On real data (Breakfast at the Frat, The Complete Journey and M5) the rules lower the world model's error on pure price cuts deeper than any it saw (0.63 to 0.56 on the Frat; 0.45 to 0.44, within chance, on The Complete Journey), but not on lasting price moves. Fourth, a small share of random test plans helps every model a little. In a generated dosing world, a stress test in a second domain, the first finding repeats, and the built-up drug level lowers the world model's error (2.86 to 1.91) though it still predicts only 0.12 to 0.32 of the true gain. The test carries to any domain where a world model is asked about actions outside its history.
67. PANO: Parallel Latent Space Neural Operator for Solving PDEs
PANO:用于求解偏微分方程的并行隐空间神经算子
AI 总结:PANO是一种新型并行隐空间神经算子,通过多并行隐空间及CAP、PaLA模块提升PDE求解性能,在9个基准上实现SOTA,显著降低单侧冲压、飞机升力系数预测误差。
链接:https://arxiv.org/abs/2611.22782
机构:University of Science and Technology Beijing(北京科技大学); Beijing Engineering Research Center of Industrial Spectrum Imaging(北京工业光谱成像工程研究中心)
作者:Zijiang Yang, Xiaomeng Wu, Chen Kuang, Dongmei Fu
英文摘要:Aggregating observations into a limited number of latent tokens and learning physical interactions in the latent space has emerged as an effective approach for Transformer-based neural operators to solve Partial Differential Equation (PDE) problems involving large-scale observations. However, existing methods typically compress observations using a single aggregation pattern, making physical information not retained by that pattern invisible to the latent space and thereby constraining physical interactions within it. To address this limitation, we propose PArallel Latent Space Neural Operator (PANO), a novel neural operator for solving PDEs that formulates physical interactions among observations as an additive combination of interaction results from parallel latent spaces with shared parameters. Specifically, PANO transforms observations into multiple distinct compressed representations through independent mappings, learns physical interactions in parallel based on these representations, and finally fuses the decoded results to predict the target physical field. Within PANO, we further propose Coupled Assignment Projection (CAP) to construct bidirectionally consistent encoding and decoding mappings, and Pairwise Latent Space Attention Layer (PaLA) to jointly update latent representations of each latent space through adaptive cross-space interactions. Extensive experiments on nine benchmarks spanning diverse physical systems, geometries, and observation scales demonstrate that PANO achieves state-of-the-art performance. Notably, PANO reduces equivalent plastic strain prediction errors on Unilateral Stamping by 28.3% (or more) and lift-coefficient prediction errors on AirCraft by 21.4% (or more) compared with existing methods. Code is available at this https URL.
68. Where Should Physics Enter? Information, Identifiability and Calibration in Probabilistic Sea Surface Temperature Forecasting
物理应在何处介入?概率海表温度预测中的信息、可识别性与校准
AI 总结:该研究探究平流-扩散先验在概率海表温度预测的异方差ConvLSTM中应介入的位置,发现学习速度无法恢复真实洋流,再分析洋流的传输设计表现不佳,校准误差是关键缺陷,修正后可提升预测性能。
链接:https://arxiv.org/abs/2611.23129
作者: Parmitha Vangapandu, Jonnakuti Pavan Kumar, Sathwik Narkedimilli, MSVPJ Sathvik, T. M. Balakrishnan Nair, Santanu Bhattacharya
英文摘要:Advection-diffusion priors are a common way to make neural sea surface temperature (SST) forecasts physics-informed. We test where such a prior must enter a heteroscedastic ConvLSTM to help, using daily NOAA OISST v2.1 for the Bay of Bengal (1982-2024) and a synthetic benchmark on the same grid with known currents of controlled strength. (i) On real data, a soft advection-diffusion residual with a learned velocity gives no reliable benefit (lead-15 CRPS -0.007 degrees C, 95% interval [-0.022, +0.008]); a plain temporal-smoothness penalty matches it, and equivalence within 3% is not established either. (ii) Learned velocities do not recover the currents, whether placed in the loss, in an explicit transport step, or in a transport step without neural correction: they correlate with the reference no better than a shuffled-date null, even on synthetic data where knowing the currents would improve day-15 forecasts by 16-23%. (iii) Supplied currents help on synthetic data, but mostly as information: giving them to the network as input channels, without transport, recovers 92% of the transport design's lead-5 gain, and a frozen-current numerical forecast does as well as either. (iv) With GLORYS12 reanalysis currents on real data, the transport design is 6% and 4% worse at leads 1 and 5; reanalysis advection explains essentially none of the daily OISST change (correlation 0.02), and frozen-current transport is worse than persistence. (v) The defect that matters is lead-dependent miscalibration shared by every model. One variance multiplier per lead, fitted on validation data, brings 90% coverage to 0.85-0.91 at all leads and lowers day-1 CRPS by 13-21%; fine-tuning on 15-day rollouts fixes long-lead coverage but costs short-lead skill. This version withdraws an earlier claim of an 81% CRPS reduction that came from an uncontrolled comparison.
13. 其他/综合机器学习 | 39 篇
69. DeltaHarness: Self-Evolving Harnesses for Resource-Adaptive Tool Execution
DeltaHarness:用于资源自适应工具执行的自演化执行框架
AI 总结:提出DeltaHarness执行引导策略搜索方法,通过QHI循环和独立确认面板自适应调整智能体执行框架,在视频编辑等任务中降低了工具执行的延迟。
链接:https://arxiv.org/abs/2611.21688
机构:University of Science and Technology of China(中国科学技术大学); Beijing Normal University(北京师范大学); Zhejiang University(浙江大学); Dmall Inc.(多点科技); East China Normal University(华东师范大学); Alibaba Group(阿里巴巴集团)
作者:Lecheng Yan, Jianze Lin, Xicheng Han, Xiantao Xu, Wenxi Li, Chenyang Lyu
英文摘要:Tool-using agents rely on execution harnesses to organize and schedule tool operations. The same task and resource description can conceal runtime conditions that reverse the latency ordering of valid execution plans. Resolving this uncertainty requires firsthand execution in the target setting. We introduce DeltaHarness, an execution-guided policy-search method that adapts agent harnesses from cold-start measurements. Its QHI loop uses the observed response to update runtime dispatch, revise physical plans, and select experiments for later proposals. An independent confirmation panel controls policy replacement. With fixed dispatch, a revised materialization policy reduces complete-delivery latency by 30.0% on one video-editing task across three fresh pairs. On held-out tasks, the evolved policies reduce latency by 8.2% relative to the best fixed plan in one seed and by 4.6% relative to the initial proposal in another.
70. AdaFlowLite: Scalable and Non-blocking Inference on Asynchronous Mobile Data
AdaFlowLite:面向异步移动数据的可扩展非阻塞推理
AI 总结:本研究针对异步移动多模态数据推理的延迟与精度问题,提出AdaFlowLite方法,采用多模态轻量级Swin Transformer实现机会性推理,显著降低延迟、提升精度并减少能耗。
链接:https://arxiv.org/abs/2611.21691
作者:Sicong Liu, Fengmin Wu, Yuan Gao, Bin Guo, Zimu Zhou, Hongkai Wen, Zhiwen Yu
英文摘要:The rise of mobile devices equipped with numerous sensors, such as LiDAR and cameras, has driven the adoption of multi-modal deep intelligence for distributed sensing tasks, such as smart cabins and driving assistance. However, the arrival time of mobile sensory data vary due to modality size and network dynamics, which can lead to delays (if waiting for slow data) or accuracy decline (if inference proceeds without waiting). Moreover, the diversity and dynamic nature of mobile systems exacerbate this challenge. In response, we present a shift to \textit{opportunistic} inference for asynchronous distributed multi-modal data, enabling inference as soon as partial data arrives. While existing methods focus on optimizing modality consistency and complementarity, known as modal affinity, they lack a \textit{computational} approach to control this affinity in open-world mobile environments. % \sysname pioneers the formulation of structured cross-modality affinity in mobile contexts using a hierarchical analysis-based normalized matrix. This approach accommodates the diversity and dynamics of modalities, generalizing across different types and numbers of inputs. Employing an multi-modal lightweight Swin Transformer (MMLST), \sysname facilitates real-time and flexible data imputation, adapting to various modalities and downstream tasks without retraining. Experiments show that \sysname significantly reduces inference latency by up to 80.4\% and enhances accuracy by up to 62.1\%, while achieving nearly a 50\% reduction in energy consumption, outperforming status quo approaches. Also, this method can enhance LLM performance to preprocess asynchronous data.
71. TIBECG: Temporal Information Bottleneck for Lead-Specific Demographic Privacy in ECG Waveform Release
TIBECG:用于心电图波形发布中导联特定人口统计隐私的时间信息瓶颈
AI 总结:本研究提出TIBECG时间信息瓶颈方法,针对心电图波形发布的导联特定人口统计隐私,在保留诊断效用的同时,优于传统变换,且支持导联特定评估。
链接:https://arxiv.org/abs/2611.21694
作者:Tsai-Min Chen (1 and 2), Jun Seita (2) ((1) Research Center for Information Technology Innovation, Academia Sinica, Taipei, Taiwan, (2) RIKEN Center for Interdisciplinary Theoretical and Mathematical Sciences (iTHEMS), RIKEN, Tokyo, Japan)
英文摘要: Transmitting an electrocardiogram (ECG) waveform can expose demographic information beyond an intended diagnostic task. We evaluate TIBECG, an architectural temporal information bottleneck, on individual leads extracted from hospital twelve-lead recordings. A 2,302,337-parameter processor varies its latent length while retaining the same nine-score consistency objective, without age or sex labels in protector fitting. Across 6,386 recordings, ten folds, 12 leads and 21 lengths, retrospective selections retained at least 95% of each lead's clean macro-F1. Their weaker age/sex interval gain exceeded the best eligible conventional transformation on all leads, by a median 22.08 percentage points. The implemented sampling-rate comparison shows that direct downsampling usually suppresses demographic inference more strongly but at a larger loss of diagnostic utility, whereas TIBECG favors utility preservation. Selection stability also varies by lead. Lower latent Kullback-Leibler (KL) divergence does not reliably rank residual sex leakage at a fixed length. These findings support lead-specific evaluation of waveform processing, not validated consumer deployment or expert-readable reconstruction. When the waveform-input diagnostic classifier is hosted as part of a proprietary cloud service rather than deployed on the device, waveform transmission is required to obtain its scores.
72. Layerwise Learning-Rate Scaling in Vision Networks: Effective-Rate Identification, Power-Mean Aggregation, and Budget Dependence
视觉网络中的分层学习率缩放:有效率识别、幂平均聚合与预算依赖性
AI 总结:本文针对从头训练的视觉网络,提出通过有效学习率识别、幂平均聚合等方法分析分层学习率缩放的效果,发现其效应随训练预算增加而衰减,可通过非训练排名提升候选方案性能。
链接:https://arxiv.org/abs/2611.21696
机构:University of Würzburg(维尔茨堡大学)
作者:Chandini Vysyaraju, Raghuvir Duvvuri, Radu Timofte, Dmitry Ignatov
英文摘要:Layerwise learning-rate scaling assigns different learning rates to different parts of a network. Although established in fine-tuning pretrained models, its effects in vision networks trained from scratch are commonly assessed by training each candidate recipe. We show that, in this setting, the rate-related component of a layerwise strategy is well summarized by an effective learning rate computable directly from its multiplier profile. Across matched-control comparisons, changes in test accuracy closely track changes in training loss (r = -0.996), which supports an optimization-progress explanation of the short-budget rate effect. Increasing the untuned base learning rate tenfold shifts the rate-matched contrast upward in all six tested architectures (p = 0.016) and increases the number of architectures in which a published discriminative schedule, excluded from model fitting, outperforms the uniform control from two to five of six (p = 0.031). Within the evaluated strategy space, a parameter-weighted power mean with exponent 0.35 (95% confidence interval [0.20, 0.50]) accounts for 42.6% of within-architecture variance, compared with 36.4% for the arithmetic mean. Ranking candidates by this scalar without training improves the selected candidate's mean percentile rank from the 26th to the 10th and reduces its mean accuracy shortfall relative to the best candidate from 3.70 to 1.12 points. The rate-related effect decays as the training budget increases; relative to tuned baselines, the overall strategy effect falls within the prespecified practical-equivalence margin in two of three architectures. A separate grouping-related contrast can persist at long budgets, although its direction and magnitude depend on the architecture.
73. Hybrid Quantum Selection of Learning Rates in Language-Model Training: Termination Criteria and an Interaction-Guided Schedule Oracle
语言模型训练中的混合量子学习率选择:终止准则与交互引导调度预言机
AI 总结:本文提出混合经典-量子框架下的量子学习率选择方法,结合Durr-Hoyer最小查找与BBHT搜索,引入交互引导调度预言机,在15亿参数语言模型微调中验证其效果,为量子叠加支持自适应学习率选择提供可行示例。
链接:https://arxiv.org/abs/2611.21701
作者:Wei-Che Hung
英文摘要:Learning-rate choices strongly affect language-model fine-tuning, but the best joint profile across layer groups is generally unknown. This paper studies quantum minimum finding for layer-group learning-rate selection within a hybrid classical-quantum training framework. It combines Durr-Hoyer minimum finding with BBHT search and compares fixed-budget and confidence-based stopping policies using a lookup oracle built from measured profile scores at fixed training checkpoints. The paper then introduces an Interaction-Guided Schedule Oracle that scores and marks joint schedules in superposition without a precomputed score table. Classical one-step loss measurements supply the main, pairwise and three-way interaction effects between layer groups. The circuit adds the effects for each schedule, compares the resulting score with the current baseline score, and marks improving schedules for amplitude amplification. Measured candidates are checked classically before acceptance. Uniform-rate training supplies the initial baseline; using schedules from classical adaptive methods as baselines for joint refinement is left for future work. In closed-loop fine-tuning of a 1.5-billion-parameter language model, each selected schedule drives the next 200 classical training steps. Benchmark accuracies lie within the range observed across five uniform-rate seeds. The results give a working example of how quantum superposition can support adaptive learning-rate selection during training; the remaining quantum limitations are discussed in the paper.
74. Margin Dynamics for Large Language Model Alignment
大语言模型对齐的边际动力学
AI 总结:本文提出LLM对齐的边际动力学理论,揭示DPO与IPO的边际特性差异,验证标签噪声的影响,并设计出TPO-S配置,其在IFEval平均得分上表现最优。
链接:https://arxiv.org/abs/2611.21766
机构:Amazon(亚马逊)
作者:Xingzi Xu, Mehmet Saygin Seyfioglu, Karim Bouyarmane
英文摘要: Practitioners choose among LLM alignment methods largely by trial and error. We provide a theoretical foundation: each pairwise method defines a vector field on the probability simplex that governs how the model redistributes mass between preferred and dispreferred completions. Projecting onto a single preferred/dispreferred pair reduces this to a one-dimensional margin ODE with a closed-form solution. These solutions reveal that DPO margins grow without bound (logarithmically as $O(\log t)$ in the gradient-flow limit) while IPO margins converge to a finite $1/(2\tau)$; that label noise induces a qualitative phase transition in DPO (unbounded $\to$ bounded) but affects IPO only quantitatively; and that gradient concentration under asymmetric weighting governs trainability at scale. We confirm all three predictions across five model families spanning four benchmarks. The closed-form solutions also enable principled method design: we prove that adding any label smoothing $\varepsilon > 0$ to SimPO creates a finite margin attractor at $\Delta^* = \beta^{-1}\log((1-\varepsilon)/\varepsilon)$, while $\varepsilon = 0$ (standard SimPO) has no attractor. We test this across three models (8B-9B) and confirm the prediction: the theory-derived configuration (TPO-S, $\varepsilon = 0.05$) is competitive with the strongest baselines, attaining the highest mean IFEval score among the compared methods on all three, validating that the predicted finite attractor is a usable operating point rather than proposing a new state-of-the-art loss.
75. VisHead-Spec: Head-Routed Visual Attention for Multimodal Speculative Decoding
VisHead-Spec:用于多模态推测解码的头路由视觉注意力
AI 总结:本文针对多模态推测解码高并发下加速效果有限的问题,提出VisHead-Spec方法,通过限制部分KV头关注图像特征,在Qwen3-VL-8B等数据集上实现了更高的吞吐量提升。
链接:https://arxiv.org/abs/2611.21839
机构:Meta Recommendation Systems (MRS)(元公司推荐系统(MRS))
作者:Xin Guo, Arushi Rai, Xiang Zhou, Qiang Zhang, Siyu Lin, Xiangjun Fan, Lizhu Zhang
英文摘要:Speculative decoding accelerates vision-language models(VLMs) while preserving the target model's outputs, but its speedup diminishes under high concurrency serving as the system becomes more compute bounded as the batch size increases. This problem is pronounced for multimodal speculative decoding as the long range of the image tokens in the prompt can significantly increase the computational cost during draft token prediction. The block-diffusion drafters (e.g. DFlash, DSpark) can effectively speed up the draft model efficiency by predicting multiple draft tokens in one forward pass, but their speedup for VLMs under high concurrency is still modest (<9%). In order to understand this gap, we examine the information flow between different modalities from target model to draft model and identify the sparsity pattern of visual attention between the target and draft models: only small set of key-value heads in the draft model need direct visual feature attention in the context and only visual features from top level layers in the target model need to be attended in the context. We therefore propose VisHead-Spec, which selects a small, fixed set of key-value (KV) heads to attend to all image features from top level layers in the target model and restricts the remaining heads to text features. On Qwen3-VL-8B and MMSpec, VisHead-Spec-8 lets 8 of the drafter's 40 KV heads read the image while keeping accepted length within 0.3% of the block-diffusion draft model baseline (i.e. DSpark). At batch size 64, its throughput gain over autoregressive decoding increases to 16.0%, compared with 9.5% for DSpark; at batch size 1, its speedup remains comparable to DSpark.
76. Inventa: Decentralized Self-Improving Agents for Autonomous Scientific Discovery
Inventa:用于自主科学发现的去中心化自我改进智能体
AI 总结:本研究提出用于自主科学发现的Inventa框架,采用自主研究单元与经验证的共享知识,在五个科学发现基准上均优于现有方法,且模型调用量显著减少。
链接:https://arxiv.org/abs/2611.21847
机构:Georgia Institute of Technology(佐治亚理工学院); Amazon(亚马逊)
作者:Prithwish Jana, Nikos Kanakaris, Purak Jain, Anoop Deoras, Sahika Genc
英文摘要:Large language model (LLM) agents can generate research ideas and improve scientific discovery through iterative experimentation. However, increasing the number of agents or iterations does not automatically translate to producing more diverse or feasible ideas. Agents may converge on similar candidates, propagate unsupported claims through shared memory components and repeatedly apply search strategies that no longer yield any improvements. We introduce Inventa, a framework for automated scientific discovery that employs autonomous research units and verified shared knowledge. Each unit creates a lineage of candidate ideas using a mutator agent. A local orchestrator agent periodically adapts the mutator's harness. When progress stalls, the orchestrator agent replaces its model or restarts the idea evolution process. Verifiers evaluate each candidate idea and record verified improvements, useful components and constraint violations in a shared knowledge, the only channel between units. We evaluate Inventa on five benchmarks spanning the research cycle, covering retrieval (ScholarCatalyst), ideation (IdeaBench), proposal writing (IdeaScientist), and scientific law discovery (EurekaBench, SciLaws-Bench). It leads all five, achieving 73.6% on EurekaBench versus 56.2% for CORAL. On IdeaBench, it discovers 26.0 novel ideas per paper, compared with 18.6 for ResearchAgent, with 82% fewer model calls.
77. tmol: a GPU-accelerated implementation of the Rosetta energy function for machine learning
tmol:用于机器学习的GPU加速Rosetta能量函数实现
AI 总结:本研究推出原生GPU的PyTorch库tmol,实现RosettaGenFF力场,可提升力场评估速度与导数集成性,批量处理吞吐量最高增362倍,用作蛋白质设计奖励函数可增加界面氢键与水桥。
链接:https://arxiv.org/abs/2611.21857
机构: Prescient Design, Genentech(Genentech公司Prescient Design部门); Institute for Protein Design, University of Washington(华盛顿大学蛋白质设计研究所); NVIDIA Corp.(英伟达公司); Department of Computer Science, University of Oxford(牛津大学计算机科学系); Department of Biochemistry, University of Washington(华盛顿大学生物化学系); Center for Advanced Therapeutics, Wistar Institute(威istar研究所高级治疗中心); Department of Bioengineering, University of Washington(华盛顿大学生物工程系); Biomedical Research Division, Korea Institute of Science and Technology(韩国科学技术研究院生物医学研究部); Sungkyunkwan University(成均馆大学); Graduate Program in Biological Physics, Structure and Design, University of Washington(华盛顿大学生物物理、结构与设计研究生项目); Graduate Program in Molecular Engineering, University of Washington(华盛顿大学分子工程研究生项目); Department of Biochemistry and Biophysics, School of Medicine, University of North Carolina at Chapel Hill(北卡罗来纳大学教堂山分校医学院生物化学与生物物理系); Lineberger Comprehensive Cancer Center, University of North Carolina at Chapel Hill(北卡罗来纳大学教堂山分校Lineberger综合癌症中心); Howard Hughes Medical Institute, University of Washington(华盛顿大学霍华德·休斯医学研究所)
作者:Andrew Leaver-Fay, Kieran Didi, Jeff Flatten, Alex Ford, Guangfeng Zhou, Nathaniel Corley, Hahnbeom Park, Maxwell W. Kazman, Rafael I. Brent, Joseph Kleinhenz, Henry Solberg, Andrew M. Watkins, Brian Kuhlman, David Baker, Frank DiMaio
英文摘要:Physics-inspired forcefields complement generative machine learning models, showing better generalization beyond the training distribution. However, these forcefields are typically slow to evaluate, and their derivatives are difficult to integrate into ML frameworks. We introduce tmol, a GPU-native PyTorch library implementing the state-of-the-art RosettaGenFF forcefield. tmol supports efficient batched differentiable scoring and whole-structure refinement with up to 362-fold throughput increase (batched). Used as a reward in generative protein design, tmol increases hydrogen bonds and water bridges at designed interfaces.
78. PromptPack: Training-Free Augmentation of Diffusion Editors for Intent-Aligned Spatial and Semantic Control
PromptPack:无需训练的扩散编辑器增强方法,实现意图对齐的空间与语义控制
AI 总结:PromptPack 是一种无需训练的扩散编辑器增强方法,通过编辑引导实现意图对齐的空间与语义控制,集成到三款编辑工具后在基准测试中表现优异,用户偏好率达 87%。
链接:https://arxiv.org/abs/2611.21865
机构:NVIDIA(英伟达)
作者:Sophia Zalewski, Mustafa Munir, Enting Liu, Anjul Patney, David Dynerman
英文摘要:Many image-editing tasks require confining changes to intended spatial regions while exactly preserving user-designated source content. Instruction-driven editing pipelines often leave such user-intent underspecified, while inpainting editors condition generation on explicitly defined edit support but not on the spatial assignment of individual edit concepts. PromptPack addresses both limitations by equipping diffusion editors with two explicit 'Edit Guides': an edit-role partition that labels each pixel as 'Keep,' 'Permit,' or 'Modify' to specify where edits may occur, and a regional-prompt assignment that maps spatial regions to text prompts to specify which concepts apply where. These guides may be manually specified or, for source-object edits, automatically derived from a text-grounded SAM 3 mask. Without retraining, PromptPack injects independently embedded regional prompts into the editor's fixed-length text sequence and uses the Edit Guides to spatially mask attention, gate flow updates, and restore protected source pixels. We integrate PromptPack into FlowEdit FLUX, FlowEdit SD3, and FLUX Kontext and demonstrate improved requested-concept placement and protected-content preservation across all three integrations on PIE-Bench++. In a pairwise user study, PromptPack achieves an 87% preference rate over FlowEdit FLUX, establishing PromptPack as a practical, training-free interface for intent-aligned diffusion editing.
79. WM-Quant: Rollout-Aware Post-Training Quantization for JEPA-Style World Models
WM-Quant:面向JEPA风格世界模型的感知回滚后训练量化
AI 总结:WM-Quant是首个针对JEPA风格世界模型的PTQ框架,通过三个组件解决回滚量化误差累积问题,在低位宽下大幅提升规划成功率,W4A4时接近全精度性能。
链接:https://arxiv.org/abs/2611.21882
机构:University of South Carolina(南卡罗来纳大学)
作者:William Hammer, Andrew Heuer, Nikhil Krishna, Matthew Grenier, Ramtin Zand
英文摘要:We present WM-Quant, to our knowledge, the first post-training quantization (PTQ) framework designed specifically for JEPA-style world models. Although world models typically feature relatively few parameters, action planning algorithms often require thousands of inferences, which can quickly become computationally expensive. PTQ is a natural way to reduce this cost. However, we show that in world models using autoregressive model predictive control (MPC) rollouts, quantization error can compound with each predictor step, resulting in action rankings from the quantized model becoming near-random. WM-Quant addresses this with three components that build on existing PTQ methods: early-encoder precision, control-path preservation, and rollout-aware scaling. Early-encoder precision protects accurate latent state and goal information to mitigate long-term error propagation. Control-path preservation allows the world model predictor to distinguish between fine-grained actions, significantly improving action rankings and execution. Rollout-aware scaling jointly calibrates encoder and per-step predictor activation ranges with respect to an entire rollout rather than a single prediction, reducing the cumulative rollout error. Across multiple world models, environments, and PTQ methods, WM-Quant substantially improves planning success at low bit-widths, approaching full-precision performance at W4A4.
80. Lost in a Few Heads: Diagnosing and Repairing Block Selection in Sparse Attention
迷失于少数注意力头:稀疏注意力中块选择的诊断与修复
AI 总结:本文针对稀疏注意力模型在固定预算下的长上下文检索精度损失,定位到特定盲注意力头,通过这些头的全注意力或4位评分器修复,可大幅提升多键检索精度,同时保留稀疏注意力的首token时间优势。
链接:https://arxiv.org/abs/2611.21917
机构:Yale University(耶鲁大学); Huawei Canada(华为加拿大)
作者:Wenhe Zhang, Boxing Chen
英文摘要: Sparse attention models read a few blocks of keys for every query at a fixed budget and can lose much of their long-context retrieval accuracy at that budget. We ask where this loss is made. Using exact attention mass selection at the shipped budget as an oracle, we show that the selector under-ranks the needle rather than running out of budget. We find attention heads that miss the needle by running a calibration probe that teacher-forces the answer and reads two quantities in every head. The first is the needle attention that the selector forfeits, and the second is the share of the answer's log-probability that fixing the head's selection restores, integrated along the path from the shipped selector to full attention. Four checkpoints of the InfLLM-V2 family and NOSA-8B lose multi-key retrieval in the same three blind heads, which attend to the needle while a mean-pooled key summary dilutes its single decisive token. Full attention in those heads or a 4-bit scorer that keeps the shipped budget recovers most of the loss without training. On MiniCPM4.1-8B at 64K, either fix lifts multi-key retrieval accuracy from 54.1 to 97.5%, level with full attention in every head. In production serving engines, full attention in those heads keeps 84 to 89% of the time to first token (TTFT) that sparse attention saves at 32K to 128K context length. Beyond the InfLLM-V2 selector lineage, one unified rule flags attention heads above a threshold swept on each checkpoint until they close most of the gap to full attention on calibration prompts. A similar small set of blind attention heads carry the loss for the SeerAttention-R gates, NSA, and DashAttention, whereas many such heads exist in MoBA. On Qwen3-8B with the SeerAttention-R gates, full attention in 18 of 288 key-value heads lifts multi-key retrieval accuracy from 29.8 to 97.7%.
81. An AI-Based OSSE Framework to Evaluate Learned Physical Consistency: A Multi-Model Study Using FourCastNet, Aurora, Prithvi WxC, and GraphCast
一种基于AI的OSSE框架用于评估学习到的物理一致性:使用FourCastNet、Aurora、Prithvi WxC和GraphCast的多模型研究
AI 总结:本研究提出首个基于AI的OSSE框架,评估FourCastNet等四种AI天气模型对观测信息的响应,发现基础模型影响最强,GraphCast存在特定数值不稳定性可通过排除100 hPa及以下层修正。
链接:https://arxiv.org/abs/2611.22024
作者:Yash Diggikar, Milton Halem, Aishwarya Raman
英文摘要:Observing System Simulation Experiments (OSSEs) are a central methodology for evaluating future atmospheric observing systems. We present the first AI-based OSSE framework to test whether different AI weather models can reproduce comparable physical impact from new observations that traditional physics based models generate. We evaluate four distinct architectural AI models on the Charney conjecture that incomplete historical observations can infer the present state of the atmospheric. We considner- FourCastNet V2, GraphCast, Aurora and Prithvi WxC and integrate each autoregressively over a 60-day forecast horizon. After a 30-day spin-up, we insert ERA5 temperature fields to infer winds and related atmospheric fields. Similar OSSE experiments are conducted for wind insertion. We measure the performance of each model's response to observational information on a 30-day data assimilation cycle forecast with and without data insertion. Distinct architectural AI weather models respond differently to the same OSSE observations. Foundation models (i.e. Prithvi WxC and Aurora), produce the strongest impact. For example, while GraphCast is the weakest, all three of the four models capture the temperature-to-geopotential height response. The two foundational models retain a positive, non-trivial, tropical zonal wind at 500 hPa response to temperature insertion. Various data insertion experiments can vary significantly even within a given architecture. We identify an architecture-specific numerical instability in GraphCast's full 37-level configuration that produces a runaway feedback concentrated at the uppermost levels with ERA5 temperature insertion correctable by excluding levels at or below 100 hPa.
82. Two Counterexamples for e-means
e-means的两个反例
AI 总结:该研究针对结合指数平均与聚类复兴的e-means提出两个反例,发现其在相同初始中心下最终损失高于标准k-means,且基于质量的复兴会破坏全局最优解,即便用k-means++初始化也会导致预期损失比值无界。
链接:https://arxiv.org/abs/2611.22065
机构:Purdue University Fort Wayne(普渡大学韦恩堡分校)
作者:Zhenisbek Assylbekov
英文摘要:e-means (Bessa et al., 2026) combines exponential averaging and cluster revival. We give two counterexamples to its published recipe with higher final loss than standard k-means (Lloyd, 1982) from identical initial centers. Mass-based revival can destroy a global optimum and make the ratio of expected losses unbounded even with k-means++ seeding.
83. TempQuant: Learning Adaptive Temperatures for Quantized Language Models
TempQuant:为量化语言模型学习自适应温度
AI 总结:TempQuant方法学习量化语言模型的自适应温度与top-p,以缩小其与全精度模型的下一个词元概率差距,在3位量化Qwen3-4B上使MATH准确率从34%提升至51%。
链接:https://arxiv.org/abs/2611.22128
机构:AMD(超威半导体公司); Cornell University(康奈尔大学)
作者:Mukund Athmaram, Liang Ma, Manoj Rao
英文摘要:Large language models are increasingly deployed at low precision, yet quantization can degrade their performance. We study how quantization interacts with decoding settings such as the sampling temperature and top-p by examining how it distorts the model's next-token probabilities. Measuring entropy during generation across deployed temperatures, we find that quantization warms the model on average, and that temperature matters much more for the performance of a quantized model than for its full-precision counterpart. Because quantization and temperature both shape the next-token distribution, we investigate how much the sampling settings can undo what quantization changes, through our method TempQuant, which learns the quantized model's temperature and top-p to bring its next-token probabilities closer to the full-precision model's. We learn these settings at two granularities. The first is a single global setting per requested temperature, which finds nearly the same temperatures that tuning on labeled answers would, without any labels. Model cards typically recommend sampling a quantized model with the same settings as its full-precision counterpart, while we find that different settings can substantially improve its accuracy. On 3-bit Qwen3-4B, for example, global TempQuant raises MATH accuracy from 34% under the configuration recommended for Qwen's quantized releases to 51%. The second sets the temperature and top-p for every token, motivated by our finding that tokens whose rankings flip under quantization tend to need warming, while tokens whose rankings are preserved tend to need cooling. When quantization degrades accuracy substantially, TempQuant recovers much of it, and when degradation is lighter, it still brings the quantized model's behavior closer to the full-precision model's.
84. An Optimization Framework for Denoising Generation
去噪生成的优化框架
AI 总结:该研究提出去噪生成的优化框架,针对多尺度高斯混合目标验证梯度上升的轨迹特性,推导源-汇恒等式与正带宽机制,通过实验证实其鲁棒性与下游泛化收益。
链接:https://arxiv.org/abs/2611.22202
机构:University of Wisconsin–Madison(威斯康星大学麦迪逊分校)
作者:Xinyu Zhou, Jiawei Zhang, Stephen J. Wright
英文摘要:Noise-unconditioned denoising learns a single score, so generation can be viewed as gradient ascent on a fixed log-density landscape. This optimization interpretation leaves two questions: do trajectories approach meaningful data, and do different initializations retain diverse outcomes? We establish both phenomena for empirical and population multi-scale Gaussian-mixture objectives. Under explicit posterior and landscape conditions, ascent converges almost surely within $O(m\sigma_{\min}^2/\tau)$ of an $m$-dimensional reference manifold with reach at least $\tau$. We derive a source--sink identity for the same fixed field and use it to prove finite-resolution regional coverage. In an affine construction, we further identify a positive-bandwidth regime where even empirical-score optimization can avoid training-sample memorization while preserving diversity at the training resolution. The same fixed objective naturally supports penalty-guided controllable generation. Experiments support our predicted mechanisms, show improved robustness under large-step guidance, and demonstrate downstream gains consistent with finite-data generative generalization.
85. Sample-efficient Personalization of Large Language Model Alignment with Multidimensional User Preference Data
利用多维用户偏好数据实现大语言模型对齐的样本高效个性化
AI 总结:本文提出MPG-DPO方法,利用自动验证器的多维评估构建个性化偏好锥指导响应选择,在匹配查询预算下,该方法在多个LLM系列上的性能优于现有样本高效偏好对齐方法。
链接:https://arxiv.org/abs/2611.22210
机构:Center for Advanced Intelligence Project, RIKEN(理化学研究所先进智能项目中心); Université de Montréal(蒙特利尔大学); Mila - Quebec AI Institute(米拉-魁北克人工智能研究所); Graduate School of Frontier Sciences, The University of Tokyo(东京大学前沿科学研究院)
作者:Zhen-Yu Zhang, Jerry Huang, Gang Niu, Takashi Ishida, Masashi Sugiyama
英文摘要:Preference alignment has become a central tool to align large language models (LLMs) with human values and desired behavior, and its online variant can further advance their capabilities beyond what static data alone offer by repeatedly generating and labeling new responses with human feedback. But costly, time-consuming annotations limit its scalability, making sample efficiency essential, and existing methods typically seek to improve it by expanding coverage of the response space or selecting pairs with the highest uncertainty for comparison. Although they are effective, these methods quantify uncertainty from the perspective of the LLM, which may differ from the uncertainty perceived by the user: a pair can appear highly uncertain to the LLM even when past comparisons from the user already imply a clear ordering. In this paper, we propose multidimensional preference-guided online direct preference optimization (MPG-DPO), which leverages low-cost multidimensional evaluations from automated verifiers on the user historical comparisons to construct a personalized preference cone guiding response selection for online preference alignment. Specifically, it first discards responses that are confidently dominated under the estimated user preference, then selects two responses that are competitive under the retained preference cone yet exhibit the strongest unresolved disagreement. Theoretical analysis shows that, with high probability, the fraction of queries spent on cone-dominated responses vanishes as learning progresses. Experiments across multiple LLM families show that MPG-DPO outperforms strong contenders under matched query budgets.
86. What Does Weight-Space Programming Add to Diffusion Editing? Program Construction versus Score-Guided Refinement
权重空间编程为扩散编辑带来了什么?程序构建与分数引导的优化
AI 总结:该研究对比权重空间编程(WSP)生成的程序与传统编辑方法在Stable Diffusion 1.4的扩散概念编辑任务中的表现,发现程序构建与分数引导优化是WSP管道的不同阶段,需分别评估。
链接:https://arxiv.org/abs/2611.22228
机构:Fudan University(复旦大学); Shanghai Institute for Mathematics and Interdisciplinary Sciences(上海数学与交叉科学研究院); Tencent(腾讯); Northwest University(西北大学)
作者:Zitian Gao, Jingyi Xu, Ziqi Han, Chengkai Lei, Ye Liu, Liangliang Shi
英文摘要:Weight-space programming (WSP) lets a language model construct executable weight updates for another network. We study diffusion concept editing through an interface in which programs define training contexts, preservation constraints and update directions, while a shared numerical fitter selects their coefficients. We compare generated programs with published editors, a shared-panel hand-written baseline and budget-matched search over conventional edits on 24 TIMED test edits for Stable Diffusion 1.4. The preregistered rewrite pipeline shows no advantage over the strong baseline. In an exploratory comparison, one-shot Astra programs improve fresh-scene VLM joint success, which requires subject presence together with whole-image target matching, by +10.0 points over the shared-panel baseline, while CLIP relative match shows no detectable change. These programs organise adjustable directions around protected text-state subspaces. Probing and rewriting the Astra programs then raises CLIP by +6.2 points while joint success changes by -9.7. On a fixed 144-image sample, five additional evaluators give point estimates in the same directions under whole-image target matching; checks of subject-local change give less decisive differences. Together, these results identify useful behaviour in complete generated programs and show that score-guided revision need not preserve it: program construction and revision should be evaluated as distinct stages of a WSP pipeline.
87. Perception into Imagination: Executable Memory for World Models
感知到想象:面向世界模型的可执行记忆
AI 总结:该研究提出可执行感知记忆(EPM)世界模型框架,在13×13 Memory Maze等任务中提升了控制性能,将持续感知与物理AI世界模型的预测及学习信号相连。
链接:https://arxiv.org/abs/2611.22294
作者:Xianyao Li, Ruitong Tian, Rui Min, Fang Xu, Jing Du
英文摘要:Physical agents must use what they previously saw or touched after visibility or contact is lost. Their actions change the implications of these remembered facts. We introduce executable perceptual memory (EPM), a world-model framework that advances explicit perceptual state alongside neural latent dynamics during imagination. Its evolving read-out informs reward prediction, value estimation and policy learning. We align reward inputs with the transitions that produced them and supervise the reward head on open-loop rollouts of recorded actions. On Memory Maze $13\times13$, EPM reaches 30.37 return at 1M interactions, compared with 23.37 when the same evolving read-out informs only policy and value on a common perceptual front end: a gain of 7.00 [3.80, 10.21]. The complete method also improves control on $9\times9$ and in MiniWorld. Probes of remembered, out-of-view arrivals show stronger reward anticipation in both visual implementations. Concentration, tactile recovery and real tactile frames demonstrate complementary roles of remembered facts in imagined-policy training, motion-dependent decisions and sensory-content retention. EPM connects persistent perception to the predicted consequences and learning signals of a world model for Physical AI.
88. Similar Is Not Useful: Selecting In-Context Demonstrations by Closed-Loop Utility
相似并非有用:通过闭环效用选择上下文演示
AI 总结:该研究针对检索增强型VLA策略,提出通过闭环效用选择上下文演示,实验表明其效果优于视觉相似性选择,在多任务及与自带演示池对比中均取得显著性能提升。
链接:https://arxiv.org/abs/2611.22295
机构:University of Texas at Austin(德克萨斯大学奥斯汀分校); Georgia Institute of Technology(佐治亚理工学院); General Intelligent Machines, Inc(通用智能机器公司)
作者:Jiachen Li, Zhigen Zhao, Shihao Li, Dongmei Chen, Long Chen, Zhuo Xu
英文摘要:Retrieval-augmented vision-language-action (VLA) policies fill their context with the demonstrations whose frames look most like the current observation under frozen DINOv2 embeddings. We ask whether visual similarity selects the demonstrations that make the policy succeed. For a frozen retrieval-augmented VLA (RICL on $\pi_0$-FAST, LIBERO), we make the demonstration pool swappable at run time and score each demonstration's utility by regressing the closed-loop success of 60 fixed-size pools on their membership. On initial states disjoint from those used for scoring, pools of the five highest-utility demonstrations succeed on 66.4% of 760 paired episodes across 19 tasks, pools of the five most similar on 55.7%, and random pools on 55.5%. Utility beats similarity by 10.8 points ($p=3.2\times10^{-6}$; positive on 15 of 18 untied tasks), including on 12 tasks the policy was never primed on, and similarity-selected pools do no better than random ($+0.1$ points). Within a task, a pool's visual similarity to the test scenes tracks its success at Spearman 0.10; across tasks, similarity separates tasks whose scenes differ and falls to chance when two tasks share a scene. The gain comes from on-task closed-loop evidence: best-of-ten direct search over pools recovers two-thirds of it with 100 rollouts. Against the shipped 20-demonstration pool, utility-curated pools of five lead by 7.0 points over 560 paired episodes from two evaluation runs ($p=0.006$), and utility-ordered retrieval inside the full pool leads by 4.5 points ($p=0.055$).
89. Fair Bubble Sort: Provably Optimal Fair Ranking with Continuous Sensitive Attributes
公平冒泡排序:具有连续敏感属性的可证最优公平排序
AI 总结:针对连续敏感属性的公平排序问题,提出基于相邻交换的公平冒泡排序算法,可证其在未加权凯梅尼距离下最优,实验显示其性能优于离散化基线。
链接:https://arxiv.org/abs/2611.22352
机构:Intesa Sanpaolo AI Research(Intesa Sanpaolo人工智能研究院); Intesa Sanpaolo Innovation Center(Intesa Sanpaolo创新中心); Intesa Sanpaolo
作者:Antonio Ferrara, Fabio Vitale, Atsushi Miyauchi, Francesco Bonchi
英文摘要:Most fair-ranking methods assume categorical protected attributes, requiring continuous attributes such as age or income to be discretized into groups. This introduces arbitrary thresholds, loses fine-grained information, and ignores the natural ordering of the attribute. In this paper, we study the problem of fair ranking with continuous protected attributes without relying on thresholding. We introduce Continuous Pairwise Parity (CPP), a fairness notion based on Kendall correlation between ranking positions and the protected attribute, and seek rankings that satisfy a prescribed CPP constraint while minimizing their Kemeny distance from an initial score-based ranking. To solve this problem, we propose Fair Bubble Sort (FBS), an efficient post-processing algorithm based on targeted adjacent swaps. We prove that FBS is exactly optimal for the unweighted Kemeny distance and provide approximation guarantees for the score-weighted setting. FBS is output-sensitive, with running time determined by the number of swaps required to meet the fairness constraint. Experiments on synthetic and real-world datasets show that FBS is scalable and consistently outperforms discretization-based baselines, achieving a superior fairness-accuracy Pareto front.
90. InfuseLoop: Cross-Trajectory Latent Injection for Scalable Parallel Looped Transformers
InfuseLoop:用于可扩展并行循环Transformer的跨轨迹隐态注入
AI 总结:该研究针对普通并行循环Transformer(PLTs)深度扩展差的问题,提出跨轨迹隐态注入机制InfuseLoop,消除训练损失峰值,提升8个循环下的困惑度与下游准确率,解码时间开销极小。
链接:https://arxiv.org/abs/2611.22426
机构:Tsinghua University(清华大学); University of California, San Diego(加利福尼亚大学圣迭戈分校); Nanjing University(南京大学)
作者:Zhaofeng Sun, Lanxiang Hu, Yu-Yang Qian, Jiaqi Leng, Yuanbo Yang, Chi Zhang, Youyou Lu, Peng Zhao, Hao Zhang
英文摘要: Parallel Looped Transformers (PLTs) use otherwise idle decoding compute by overlapping recurrent iterations. However, vanilla PLTs scale poorly with depth: they exhibit training-loss spikes, and their quality gains shrink as more loops are added. The benefit of additional loops depends on the information available to each iteration, so we examine whether the existing computational dependencies limit effective depth scaling. We find that only the preceding in-trajectory state is passed across loop boundaries, even though completed cross-trajectory representations are already available. Based on this finding, we propose InfuseLoop, a cross-trajectory latent injection mechanism. It aggregates causally available hidden states from later prediction trajectories through lightweight attention residuals. This design preserves token-parallel training and concurrent loop execution, and requires only $O(Rd)$ additional hidden-state storage. Experiments show that InfuseLoop eliminates training-loss spikes and improves validation-loss scaling up to eight loops. At eight loops, InfuseLoop reduces perplexity by 3.4--17.1% relative to vanilla PLTs and by 10.5--37.3% relative to the non-looped baseline. It also improves average downstream accuracy by 0.84 and 1.99 percentage points over these two baselines. InfuseLoop incurs only 0.3% decoding time over vanilla PLTs and 23.3% over the non-looped baseline, while keeping the same logical KV-cache footprint as vanilla PLTs.
91. How Many Training Labels Are Wrong? Auditing a Fitted Model through Gradient Allocation
有多少训练标签是错误的?通过梯度分配审计拟合后的模型
AI 总结:本研究提出通过梯度分配审计拟合后的模型,可准确判断训练集错误标签数量与位置,在CIFAR系列数据集上表现优于常规方法,且无需额外训练或干净数据。
链接:https://arxiv.org/abs/2611.22449
机构:The University of Sydney(悉尼大学)
作者:Yuchen Li, Zongqi Fan, Xintong Li, Nguyen H. Tran, Ken-Tye Yong
英文摘要:How many labels in a training set are wrong, and which ones? The usual answers read a held-out set or the model's outputs. A held-out reading depends on how the held-out set was drawn, and an output-based reading fails once the model has fitted its wrong labels, because the loss of a fitted wrong label is small and the prediction agrees with it. We read the answer from the per-example gradients of a single trained model. Each example, together with its label, asks the network to change in a particular way. We normalise the magnitudes of its gradient into a gradient allocation and compare it with the mean allocation of every class. An example is flagged when relabelling it as its nearest other class makes its allocation fit better than under its own label, and the share of flagged examples is the reading of the noise rate. The rule has no threshold and needs no clean data, no held-out set and no further training. After the model has fitted 99-100% of its wrong labels, the allocation still separates wrong from right labels with an AUC of 0.94-0.96 on the noisiest human labels of CIFAR-10N, where a feature-space margin falls to 0.79-0.84 and the loss, the gradient norm and the self-influence of an example to 0.73-0.76. On such a fitted model, readings of the noise rate from predictions or feature margins fall to 0-6% at true rates of 8-58% on CIFAR-10 with ViT-S, while the audit reads 54-96% of the true rate, and at least 93% of the examples it counts carry wrong labels. On a briefly trained model, before the wrong labels are fitted, the reading lies within 1.7 percentage points of the true noise rate on CIFAR-10 from 0% to 58% noise. Confident learning, which trains four further models, separates wrong labels about as well (AUC 0.925-0.998 against 0.900-0.995) and mostly reads the noise rate more closely, at a lower precision. An audit of 5,000 examples takes 14 to 27 seconds on one GPU.
92. Escaping Latent Traps by Remembering: Training-Free Episodic Kernel Repulsion for World-Model Planning
通过记忆逃离潜在陷阱:用于世界模型规划的无训练情景核排斥方法
AI 总结:本文针对世界模型规划中智能体易陷入潜在陷阱的问题,提出无训练的情景核排斥方法,在多个模型和基准测试中显著提升了规划成功率,且计算效率高、可迁移性强。
链接:https://arxiv.org/abs/2611.22470
机构:Institute of Integrated Circuits and Artificial Intelligence, Qiantang Institute for Advanced Study(钱塘高等研究院集成电路与人工智能研究所); Zhejiang University(浙江大学); Guangdong University of Technology(广东工业大学)
作者:Yican Zhao, Bolin Yang, Ruizhe Wang, Haiyang Guo, Fuji Yang, Bingo Wing-Kuen Ling, Zelin Zang
英文摘要:Latent world models plan towards a goal image by minimising the distance between predicted and goal latents. This distance is not geodesic: it has local minima, latent traps, behind walls, at corners and in dead ends, and memoryless receding-horizon replanning keeps the agent in them, revisiting states far from the goal. We propose Episodic Kernel Repulsion (EKR), a training-free cost term that penalises candidate plans by a top-$k$ Gaussian kernel to a novelty-gated memory of the latents the agent has actually visited, rescaled by the current base cost so that its strength is comparable across models. Across two JEPA world models and four benchmarks, EKR raises average success from 31.8% to 46.7% (LeWM) and from 54.7% to 64.0% (H-JEPA) with the same model, planner and budget. Same-memory counterfactuals show that, in most settings, EKR pushes plans away from visited states, in navigation, often at a higher base cost; stuck episodes and revisits drop accordingly. In seven of eight settings it matches or exceeds $4\times$ samples at $0.8$--$1.5\times$ compute. Without tuning, it transfers to DINO-WM with CEM: $+6.7$ and $+19.3$ points on official and longer-goal Wall tasks, and from 44.7% to 100.0% and from 21.0% to 32.0% on long-horizon PointMaze and Push-T.
93. Right Labels, Wrong Reasons: Signed Feature Attribution for Multi-Label Explainability
正确标签,错误理由:面向多标签可解释性的带符号特征归因
AI 总结:针对多标签分类器可能因错误理由预测正确标签的问题,提出FLEX全局代理模型,其能生成带符号特征归因列表,在多数数据集上表现优于逐标签方案,可标记错误理由预测的标签。
链接:https://arxiv.org/abs/2611.22515
机构:AC Online Algonquin College(阿尔冈昆学院AC在线)
作者:Kenton White
英文摘要: When a multi-label classifier assigns the right labels, it may do so for the wrong reasons. Diffuse feature combinations may vary across classifier families and lack a dominant feature. This paper proposes FLEX (Fisher LDA Explainer), a global surrogate that fits a single direct multi-label Fisher LDA over the full binary label matrix, producing signed feature attribution lists for each label in a single coherent coordinate system. Across the evaluated benchmarks and classifier families, FLEX reproduces the black box's predictions with consistently high fidelity. Pooling covariance across labels outperforms a per-label alternative on most datasets, with the advantage growing largest where labels are numerous and positive examples are scarce. FLEX's fidelity-complexity Pareto curve is monotonically increasing, giving practitioners a predictable path to higher-fidelity explanations as more features are used, and its in-sample fidelity matches or leads Binary Relevance alternatives once a modest feature budget is allowed. The magnitude of each discriminant weight vector bounds its label's separability, flagging labels that may be predicted correctly for the wrong reasons. The evaluated classifiers largely agree on which predictions to make, but reach those decisions through different feature subsets when no single feature dominates.
94. AlignedHPO: Efficient Surrogate-Guided Loss Hyperparameter Optimization
AlignedHPO:高效的代理引导损失超参数优化
AI 总结:针对深度学习损失超参数调优的低效问题,提出AlignedHPO框架,通过梯度对齐与代理验证目标,将超梯度计算提速近12倍,在两类任务中性能优于或媲美隐式微分。
链接:https://arxiv.org/abs/2611.22658
机构:Sber AI Lab(Sber AI实验室); HSE University(高等经济大学); ISP RAS Research Center for Trusted Artificial Intelligence(俄罗斯科学院信息学与自动化研究所可信人工智能研究中心)
作者:Ivan Karpukhin, Andrey Savchenko
英文摘要:Modern deep learning losses involve multiple hyperparameters, such as weights, scales, and margins, typically tuned manually or through expensive black-box optimization despite their differentiability. We propose AlignedHPO, an efficient framework for tuning differentiable loss hyperparameters using gradient alignment and surrogate validation objectives. Our method employs Jacobian-vector products to reduce hypergradient computation time and memory, while differentiable metric surrogates guide optimization toward task-specific evaluation criteria. We address gradient magnitude ambiguity and stochastic instability through normalized alignment and gradient smoothing. Experiments with generalized Focal Loss and ArcFace demonstrate effective adaptation across long-tailed classification and open-set face recognition. AlignedHPO achieves competitive or improved performance compared to implicit differentiation, with nearly 12x faster hypergradient computation. These results highlight the potential of efficient surrogate-guided loss hyperparameter optimization.
95. CULypso: A Clustered Federated Unlearning paradigm for Vehicular Ad-Hoc Networks
CULypso:面向车载自组织网络的聚类联邦遗忘范式
AI 总结:针对车载自组织网络的隐私保护需求,提出CULypso聚类联邦遗忘算法,实验显示其在时间、存储负载等指标上比FedEraser优约65%,优于完全重训练基线。
链接:https://arxiv.org/abs/2611.22682
作者:Eleftheria Chini, George Margaritis, Dimitrios Katsaros
英文摘要:Unlearning has emerged as an essential security and privacy-preserving solution in Machine Learning, enabling networks to support their clients' rights for data erasure, as mandated by data protection regulations. Introducing such a non-trivial task to an edge, highly mobile and time-sensitive vehicular network is an even greater, largely unexplored challenge. This article argues that a hierarchical Federated Unlearning approach could be aptly integrated into vehicular ad-hoc networks (VANETs), leveraging topology and movement to reduce communication, resource and time costs, while effectively forgetting a client's influence on the global model updates. In order to achieve this, we propose CULypso, an approximate, surgical, hierarchical unlearning algorithm tailored for clustered VANETs. Inspired by the nymph in Greek mythology, Calypso, whose name translates to 'she who conceals the knowledge', CULypso, with the capital letters referencing \textbf{C}lustered \textbf{U}n\textbf{L}earning, aims to achieve exactly what its name suggests: conceal, erase or forget in our case, knowledge it contains. Extensive experiments uphold the potency of our paradigm with it performing approximately 65\% better than FedEraser in terms of time, server storage load, and unlearning fidelity while utterly outperforming baseline full retraining.
96. HADVQ: Hessian Aware Directional Vector Quantization
HADVQ:海森感知方向向量量化
AI 总结:针对大语言模型超低比特量化的挑战,提出HADVQ方法,通过随机预处理、极坐标解耦量化、E₈格码本及代理海森保留重要通道,提升量化后模型准确率。
链接:https://arxiv.org/abs/2611.22721
机构:Applied AI Institute(应用人工智能研究所); Luzin Research Center(卢津研究中心)
作者:Daniil Li, Mars Davletshin, Vadim Klyzhenko, Artem Danilishin, Valery Sandul, Daniil Ryazanov, Ekaterina Muravleva
英文摘要:Ultra-low-bit quantization of large language models is a challenging task because aggressive compression severely distorts model parameters and disproportionately harms a small number of highly sensitive channels. We present a vector quantization method tailored for this regime. Our approach begins with random preconditioning, which regularizes the weight space and makes it more amenable to quantization. We then represent weights in polar coordinates and decouple the quantization of directions and magnitudes. To quantize directions, we leverage the structure induced by random preconditioning and design a structured codebook based on the $E_8$ lattice, yielding a compact yet expressive representation. To further reduce degradation, we detect extremely important channels using a proxy Hessian and preserve them with an additional cost of only 0.16 bits. Together, these components provide an effective quantization pipeline for LLMs in the extremely low-bit regime. Empirical results demonstrate improved accuracy, highlighting the benefit of combining geometric structure, lattice-based codebooks, and selective channel preservation.
97. Virtual Double Oracle: Faster Convergence by Leveraging Strategies That Were Watching All Along
虚拟双 oracle(VDO):利用一直被观测的策略实现更快收敛
AI 总结:该研究针对两人零和博弈,提出虚拟双 Oracle(VDO)及其折扣变体 dVDO,通过利用策略的时间平均对手效用初始化新策略,降低可利用性,在矩阵博弈、扑克等任务上表现优于现有方法。
链接:https://arxiv.org/abs/2611.22806
机构:London School of Economics(伦敦政治经济学院)
作者:Ruilan Wang, Francisco Aristi Reina, Katerina Papadaki
英文摘要:Double Oracle (DO) and PSRO approximate Nash equilibria in two-player zero-sum games by iteratively expanding a strategy population with best responses supplied by an oracle. Online Double Oracle (ODO) interleaves multiplicative weights updates (MWU) with population discovery and obtains an $O(\sqrt{k\ln k/T})$ rate, where $T$ is the number of learning rounds and $k$ is the larger of the two players' final population sizes. We identify the reinitialisation step as an underexplored design choice: at each expansion, MWU-based methods reset newly discovered strategies to a uniform prior. We introduce Virtual Double Oracle (VDO), which exploits a bilinear identity (a strategy's full hindsight loss history collapses to its utility against the time-averaged opponent) to initialise each new strategy with the MWU weight it would have had from the start. We prove that VDO achieves $O(\sqrt{k/T})$ exploitability over the discovered population as it grows, provided the cumulative cost of adding strategies is at most linear in $k$, a condition supported by our empirical measurements. Discounted VDO (dVDO) moderates historical preferences to reduce over-concentration at reinitialisation. Experiments across matrix games, poker, and Goofspiel show that, with a suitable discount, dVDO can substantially reduce finite-horizon exploitability relative to both VDO and uniform-reset ODO, and can improve the discovered strategy populations.
98. When the tail is earned by its companions: measuring label dependence in long-tailed multi-label classification
当尾部标签由其同伴标签赋予时:长尾多标签分类中的标签依赖性测量
AI 总结:该研究针对长尾多标签分类中分类器依赖同伴标签恢复稀有尾部标签的问题,提出标签内估计量等方法,发现宏F1认可的尾部真阳性存在18.4%的依赖损失,且架构会影响该依赖性。
链接:https://arxiv.org/abs/2611.22859
机构:School of Computer Engineering, Pontificia Universidad Católica de Valparaíso(瓦尔帕莱索天主教大学计算机工程学院)
作者:Christoffer Löffler
英文摘要:A rare label seldom occurs alone, and a classifier can recover it from the frequent labels it co-occurs with, its companions, without having learned the label itself. Existing diagnostics of this need a trained model: none says before training whether a corpus admits the failure, or what the standard aggregate hides. Macro-F1, like any measure built from per-label confusion counts, is unchanged by how a label's recall divides between occurrences with and without a companion. We introduce a within-label estimand of that division, a reading rule under which a fall in dependence counts only if companion-free recall rises, and a statistic of training labels that bounds, before training, how far reported recall can exceed companion-free recall. Across 23 corpora, 7 training objectives and five families of label-correlation mechanisms, companion-free recall falls short on all six corpora where the estimand is defined. Up to 18.4% of the tail true positives macro-F1 credits would vanish at the companion-free rate, within the training-label bound on every corpus. No long-tail objective reduces the dependence, but architecture moves it. Models whose tail score is computed from other labels' predictions are strongly dependent.
99. Beyond Attribute-Centric Fairness: Modelling and Mitigating Unified Bias Representation in Omni-modal Large Language Models
超越以属性为中心的公平性:全模态大语言模型中统一偏差表示的建模与缓解
AI 总结:针对全模态大语言模型(OLLMs)的属性中心去偏范式缺陷,提出OmniDebias方法,通过解耦偏差与合法知识、构建统一偏差表示,提升OLLMs跨属性与模态的公平性且保持良好效用。
链接:https://arxiv.org/abs/2611.22889
机构:Univeristy of Oxford(牛津大学); The Chinese University of Hong Kong(香港中文大学); The University of Sydney(悉尼大学)
作者:Runqi Lin, Suqin Yuan, Zihao Fu, Chris Russell
英文摘要:Omni-modal large language models (OLLMs) provide joint multimodal understanding and generation capabilities, making them well-suited to complex environments and promising backbones for next-generation AI. Despite these advances, the harmful stereotypes encoded in such models manifest in increasingly diverse forms, leading to broader and more severe societal consequences. Existing debiasing methods typically impose separate fairness objectives for each sensitive attribute and modality to minimise demographic-dependent variations. However, in omni-modal settings, this attribute-centric paradigm struggles to jointly address diverse social biases with distinct fairness objectives and to generalise beyond predefined attributes. To this end, we propose OmniDebias, one of the first methods for promoting fairness in OLLMs, which employs a bias-centric formulation to extract and mitigate a shared bias representation across attributes and modalities while preserving attribute-related knowledge. Specifically, we exploit activation inconsistencies of a given attribute (e.g., gender) across semantic contexts (e.g., occupations) to disentangle bias from legitimate knowledge. We further consolidate multiple extracted biases into a shared representation, formulating a unified objective that captures the underlying structure across social biases and generalises to unseen attributes. Extensive experiments demonstrate that eliminating unified bias with both post-hoc editing and fine-tuning can consistently improve fairness across diverse attributes and modalities in OLLMs while maintaining favourable utility.
100. ResoLUT: Distilling Residual Networks into Lookup-Table Networks
ResoLUT:将残差网络蒸馏为查找表网络
AI 总结:ResoLUT是首个保留ResNet多比特残差流的卷积逻辑网络,通过将ResNet逐层蒸馏为LUT树,在CIFAR-10、CIFAR-100等数据集上取得最优结果,且精简模型可在FPGA上实现高吞吐量。
链接:https://arxiv.org/abs/2611.22955
机构:Transcendent Logic(超越逻辑公司); Zurich University of Applied Sciences (ZHAW)(苏黎世应用科技大学); Wroclaw University of Science and Technology(弗罗茨瓦夫科技大学); Vilnius University(维尔纽斯大学)
作者:Katarzyna Fojcik, Aidas Mikalauskas, Matthias Kampa, Renaldas Zioma, Jogundas Armaitis
英文摘要: Logic networks built from Boolean gates or lookup tables (LUTs) promise efficient hardware inference, but training them at depth and learning their connectivity remain difficult. We do not train logic as logic. Instead, ResoLUT distills a trained residual network (ResNet) layer by layer into LUT trees, preserving its multi-bit residual stream and BatchNorm while extracting connectivity from the teacher rather than learning it. To our knowledge, ResoLUT is the first convolutional logic network in computer vision to retain a ResNet-style multi-bit residual stream. Hard LUT decisions are used throughout training and inference, avoiding a discretization gap in the logic. ResoLUT-20 reaches 95.66% on CIFAR-10 with twice the trees, and ResoLUT-56 reaches 78.87% on CIFAR-100 and 53.69% on ImageNet at 64px, the best results of logic and lookup-table networks on CIFAR-10 and CIFAR-100, showing that LUT networks can scale in depth. We explore and quantify the role of the residuals and the tree budget and shape. The most optimized model, a ResoLUT-8 with one eighth of the trees and further optimizations, fits on an FPGA and reaches 90.04% at 189k images per second, the highest throughput among the compared models above 85% on CIFAR-10.
101. DDSL: Iterative Self-Labeling for Financial Market States
DDSL:面向金融市场状态的迭代自标注方法
AI 总结:本研究针对金融市场状态标注问题,提出迭代自标注框架DDSL,通过耦合维特比标注器与Transformer分类器提升标注质量,在多股指数据集上验证了其性能增益与可学习性可控性。
链接:https://arxiv.org/abs/2611.22958
机构:New York University Shanghai(上海纽约大学)
作者:Xianbin Gu, Qingyue Zhu, Yuhong Geng
英文摘要:Financial market states lack a unique ground truth and must therefore be constructed from price data. Conventional labeling methods typically define states independently of the downstream classifier, leaving unresolved whether the resulting labels can be recovered from information available at prediction time. We propose DDSL, an iterative self-labeling framework that couples a price- and grammar-guided Viterbi labeler with a Transformer classifier through model feedback. DDSL constructs a nine-state representation that jointly encodes current and future market trends while retaining structured price emissions and temporal constraints throughout refinement. Experiments on SSE50, CSI300, CSI500, and NASDAQ100 show that DDSL increases held-out joint-state macro F1 by 78%-96% over single-pass dynamic-programming labeling. An independent GRU classifier that never participates in the labeling loop reproduces the same trend, while label entropy and class coverage remain stable, indicating that the gain transfers across architectures and is not caused by label collapse. We further show that learnability is controlled by both iteration and the model-feedback weight $\beta_{\mathrm{model}}$. Iterative gains diminish as successive label systems converge, whereas stronger feedback increasingly widens the performance gap between the feedback Transformer and the independent GRU. Increased learnability is accompanied by declining segment-level price-feature separation after an initial improvement. Finally, a non-anticipative event study provides partial external validation: refinement strengthens a conditional BEAR$\rightarrow$BULL reversal signal on the three A-share datasets, while the corresponding effect has the opposite sign on NASDAQ100. These results position DDSL as a structured framework for constructing market-state labels whose learnability can be explicitly controlled and evaluated.
102. NVIDIA Kumo Tabular: Pushing the Accuracy$\unicode{8211}$Efficiency Frontier for Tabular Prediction
NVIDIA Kumo Tabular:突破表格预测的精度-效率前沿
AI 总结:NVIDIA推出Kumo Tabular系列表格基础模型,含三种规模,在八项表格预测基准中表现最优,精度随规模提升,且预测速度优于同类模型,相关权重与代码已开源。
链接:https://arxiv.org/abs/2611.23001
机构:NVIDIA(英伟达)
作者:Jingang Qu, Valter Hudovernik, Martin Jurkovič, Cedric Lorenz, David Holzmüller, Federico Lopez, Dmitry Gordeev, Akihiro Nitta, Ramona Bendias, Ardrian Wong, Vedaant Jain, Vignesh Kothapalli, Gilberto Titericz Junior, Jure Leskovec, Matthias Fey
英文摘要:We introduce Kumo Tabular, a family of open tabular foundation models for classification and regression, in three sizes: Kumo Tabular-S (27M parameters), Kumo Tabular-M (61M), and Kumo Tabular-L (214M). Pretrained only on artificial tables from the Kumo prior, they can make predictions on new tables without any training. In our approach, a table encoder compresses each row into an embedding, and an in-context learning transformer predicts the unlabeled rows from the labeled ones. Kumo Tabular introduces architectural innovations that improve both training stability and predictive performance, enabling the model to scale up to 150M cells on modern GPUs. Kumo Tabular-L ranks first on eight benchmarks: TabArena, TALENT, BeyondArena, ScoringBench, STRABLE, MulTaBench, RamanBench, and TabBench-Bio. These cover probabilistic regression, tables with over 1M rows or 100K columns, and tables with strings, text, or images. Accuracy improves with model size in all tested comparisons. All three sizes lie on the accuracy-speed frontiers of TabArena and BeyondArena. Kumo Tabular-L predicts faster than the runners-up, LimiX-2 on TabArena (36 times) and TabPFN-3.5 on BeyondArena (1.7 times). Weights and pretraining code are released under OpenMDW-1.1 and Apache 2.0, respectively.
103. Pareto-Informed Smart "Predict, then Optimize" for Multi-Objective Combinatorial Problems
面向多目标组合问题的帕累托知情智能“预测后优化”方法
AI 总结:针对多目标组合优化问题,提出帕累托感知的SPO+扩展方法,通过利用帕累托前沿设计损失函数,在多目标最短路径、背包问题上优于传统基线与标量化SPO+方法。
链接:https://arxiv.org/abs/2611.23076
机构:University of Würzburg(维尔茨堡大学)
作者:Aabhash Dhakal, Tim Lachner
英文摘要: Many real-world planning applications, from logistics, routing to resource allocation, give rise to optimization problems that need to balance multiple, conflicting objectives. While machine learning is often used to predict the unknown parameters or costs of these problems, standard models trained on prediction accuracy can lead to suboptimal plans. Decision-Focused Learning (DFL) addresses this by integrating the downstream optimization task into the training objective. However, existing DFL methods for multi-objective problems rely on simple scalarization, which fails to capture the rich trade-offs of the Pareto Frontier. To address this gap, we propose a Pareto-aware extension to single-objective Smart "Predict, then Optimize" (SPO+) for multi-objective combinatorial optimization problems. Our loss function leverages the true Pareto Frontier, forming a target decision that aligns with a sampled preference vector, and minimizes regret against this target. Experiments on multi-objective shortest path and multi-objective knapsack problems show that our approach outperforms traditional two-stage Mean Squared Error (MSE) baselines. It also surpasses scalarization-based SPO+ approaches.
104. The Illusion of Grounding: Do Vision-Language Models Weigh Retrieved Text and Images Like an Ideal Observer?
grounding的错觉:视觉语言模型是否像理想观察者一样权衡检索到的文本与图像?
AI 总结:该研究构建GIB-200K数据集,提出知识惯性等指标,评估4个VLMs发现模型权衡模态的方式存在差异,部分模型过度依赖文本,且误导下准确率大幅下降。
链接:https://arxiv.org/abs/2611.23127
机构:eClerx(埃克莱克斯公司)
作者:Basavaraj R M, Harshal Chandel
英文摘要:Retrieval-augmented vision-language models (VLMs) answer from two sources of evidence that can each fail: the retrieved text can be wrong or noisy, and the image can be degraded or contradicted. We introduce GIB-200K, 200,000 image-question items built from COCO and Visual Genome, each expanded into controlled stress conditions: hard-negative retrieval noise (1-20 documents), visual ambiguity (Gaussian and motion blur, low-resource crops) and cross-modal conflict with counterfactual flip pairs. Instead of only measuring degradation, we ask how models combine the two modalities. From ideal Bayesian cue combination we derive a testable prediction that yields formal definitions of Knowledge Inertia (text over-weighted relative to the model's own image evidence) and Sensory Capture (the reverse), and we estimate the weights with a measurement-error-corrected regression whose recovery of known weights we verify. A counterfactual flip decomposition separates modality bias from concept priors, and a Grounding Illusion Score measures how much clean accuracy is not grounded in the image. We evaluate 4 open models on one frozen, seeded evaluation split of 2,000 GIB-200K items. All models are accurate on clean inputs (92.5%-94.2%) but adverse accuracy falls to 41.3%-59.9%. The two LLaVA models over-weight the text (Knowledge Inertia, w_T/w_I = 1.229 and 1.322), whereas InternVL3 favours the image (w_T/w_I = 0.680); between 34.8% and 71.4% of the clean answers that are correct are abandoned when the document lies. A training-free consistency check that compares the answer with and without retrieval trades coverage for accuracy under misleading documents, while falling back to the image fails when the image is heavily degraded. All per-instance outputs, code, and an audit that recomputes every reported open-model number are released.
105. Learned versus Kinematic Trajectory Prediction under Reduced Observation Frequency
低观测频率下的学习型轨迹预测与运动学轨迹预测对比
AI 总结:本研究在Argoverse 2数据集上对比不同频率下学习型RNN与运动学CV、CA模型的轨迹预测性能,发现低频率下学习型模型仍有优势,1Hz时插值的10Hz模型准确率下降11.5%,假设仅部分成立。
链接:https://arxiv.org/abs/2611.23198
作者:Nukhbala Gadzhiev, Alexander Morochko
英文摘要:Trajectory prediction models for automated driving are usually trained on data with a fixed sampling frequency, while positions received via vehicle-to-everything (V2X) communication arrive at 1-10 Hz. We study whether a learned predictor that uses only the observation history keeps its advantage over constant velocity (CV) and constant acceleration (CA) models when the history frequency drops. On the Argoverse 2 motion forecasting dataset, the 5 s history of the focal agent was downsampled from 10 Hz to 5, 2, and 1 Hz, and the 6 s future was predicted without maps or neighboring agents. An LSTM-based recurrent neural network (RNN) trained for each frequency was compared with the same network trained at 10 Hz and applied to sparse histories after linear interpolation. Hypotheses and thresholds were fixed before training, the main RNNs were trained with ten seeds each, and results on 24,984 test scenarios have bootstrap confidence intervals over scenarios and seeds. The average displacement error of the frequency-specific RNN grew by 3.1% at 5 Hz, 10.1% at 2 Hz, and 20.3% at 1 Hz. Contrary to our hypothesis, its advantage over the better kinematic model did not vanish on straight motion and was 24-30% on both straight motion and turns at every frequency, while for pedestrians CV was more accurate. The 10 Hz model with interpolation was at most 1.3% less accurate than the frequency-specific model down to 2 Hz but 11.5% less accurate at 1 Hz. Each hypothesis was only partly supported.
106. Distortion-Free Semantic Watermarks for Black-Box Generative Models
黑盒生成模型的无失真语义水印
AI 总结:本研究提出一种仅需黑盒访问的通用无失真语义水印算法,针对图像生成训练专用编码器,在开源与闭源模型上验证了其对内容保留型变换的鲁棒检测能力。
链接:https://arxiv.org/abs/2611.23208
机构:Tel Aviv University(特拉维夫大学)
作者:Aviv Yossef, Or Patashnik, Or Zamir
英文摘要: We present a provably distortion-free semantic watermarking algorithm that requires only black-box sampling access to a generative model and black-box access to a semantic embedding. A watermark is called distortion-free if, averaged over its secret detection key, it leaves the generator's output distribution unchanged; this provides a formal guarantee that watermarking does not reduce output quality. Conventional such watermarks are often brittle to transformations. This is especially problematic for images, where benign operations such as compression, cropping, or resizing change the entire pixel representation. Semantic watermarks therefore require detection to be robust to content-preserving transformations. Previous methods combining these properties rely heavily on specific white-box properties of the underlying model: either to autoregressive text generation or to diffusion-specific access, such as control of the initial noise or diffusion inversion. In contrast, our construction is fully generic: It only interacts with the generative model and the semantic embedding, a function that determines the notion of "preserving content" under which detection should be robust, via sample and query access. We then instantiate this framework for image generation. We characterize the two competing properties required of the semantic embedding: it should remain stable under transformations that preserve an image's content, but should distinguish independent generations from the same prompt. We train an image encoder directly for this tradeoff. The resulting encoder improves separation while maintaining stability over off-the-shelf alternatives and enables strong aggregate detection while remaining robust to various transformations. We demonstrate the effectiveness of our watermark scheme on both open-source and closed models, which is possible only due to restriction to black-box access.
107. Do LLM Agents Still Need to Be Told How to Self-Improve?
大语言模型智能体仍需被告知如何自我提升吗?
AI 总结:本文研究自主范式下大语言模型智能体的自我提升能力,发现足够能力的模型可在无外部指定优化方式时自我提升,且学习成果可迁移至多类游戏基准。
链接:https://arxiv.org/abs/2611.23219
机构:MIT CSAIL(麻省理工学院计算机科学与人工智能实验室); University of Tübingen(蒂宾根大学)
作者:Shobhita Sundaram, Cansu Sancaktar, Phillip Isola
英文摘要:LLM agents increasingly drive their own learning. Self-improvement methods delegate various components of model optimization to the models themselves. As models become more capable, this motivates a basic question: do these increasingly capable models still need us to specify how they should self-improve? In this paper we study an autonomous paradigm in which, given basic primitives, resources, and a general goal for self-improvement, agents construct their own self-learning algorithm for context-based optimization. Using the setting of gameplay as a testbed, a frozen LLM agent is given a compute budget and tools for collecting experience and updating text documents, and allowed to freely experiment without an externally imposed loop or access to real-world data. We find that sufficiently capable models do self-improve under these conditions, demonstrating strategies that align with classic cognitive frameworks for self-regulated learning and constructs from machine learning literature on self-generated curricula. Furthermore, we study the extent to which the learned text artifacts transfer to external benchmarks, observing improvement across different game horizons, mechanics, and 3D navigation.