机器学习学术速递[7.14]

AI 总结：研究提出低秩注意力残差（LR-AttnRes），包括投影式和切片式，通过使用低维键进行路由，解耦路由与残差内容，减少运算量并提升性能，证明深度路由在较少维度下也有效，还发布了代码和模型。

链接：https://arxiv.org/abs/2607.09694

作者：Jonathan Su

英文摘要：Attention Residuals replace the fixed residual sum with depthwise attention over previous sub-layer outputs in large language models (LLMs), but use each output as both a full-dimensional key and value. This couples routing with representation and makes depth-routing scores scale with the hidden width $d$. We propose Low-Rank Attention Residuals (LR-AttnRes), which keep full-dimensional residual values while using $r$-dimensional keys, with $r \ll d$, for routing. Projected LR-AttnRes emits learned low-rank keys from existing output projections, decoupling routing from residual content and achieving the best validation loss among the variants tested. Sliced LR-AttnRes uses the last $r$ dimensions of each value as the routing key, removing the auxiliary key-projection path and reducing residual-side FLOPs while still improving performance. Comprehensive sweeps show that depthwise routing can be effective with far fewer dimensions than the model width. We release code and models to facilitate future research.

AI 总结：研究低精度训练中权重更新低于半ULP时坐标冻结问题，通过高精度轨迹和目标尾数长度预测冻结，以小型GPT、GPT - 2变压器等为例展示结果，还表明随机舍入可消除冻结，此条件在多种模型适用。

链接：https://arxiv.org/abs/2607.09800

机构：University of Illinois at Urbana-Champaign（伊利诺伊大学厄巴纳-香槟分校）

作者：Zekai Shang

英文摘要：Training in reduced floating-point precision can silently halt learning: when a gradient-descent weight update falls below half the unit in the last place (ULP) of the weight, it rounds away and that coordinate freezes while its gradient is still nonzero. The freeze is deterministic, governed by a per-coordinate half-ULP condition, and predictable from a high-precision trajectory and the target mantissa length alone, without low-precision data. In a small GPT trained under the standard AdamW-plus-cosine recipe with bf16-equivalent stored weights, training proceeds normally and then permanently freezes just past mid-run, within four steps of the a-priori prediction. In a $124$-million-parameter GPT-2 transformer whose weights are constrained to the $8$-bit floating-point grid after every optimizer step, with no master weights, the dense weights freeze at initialization in both fp8 formats -- predicted \emph{a priori} from an fp32 reference -- and validation loss plateaus while full precision keeps improving. Stochastic rounding removes the persistent freeze, and the same reference predicts that too. The condition transfers across frozen-feature regression, a mantissa-truncation emulator spanning $128\times$ in precision, small networks, and a CNN on MNIST: a computable axis of low-precision training, not diffuse noise.

AI 总结：研究固定状态序列模型与注意力模型间的中间地带，提出基于DP均值聚类规则的可学习稀疏缓存，有静态和惊喜自适应两种形式，在基准测试中表现良好，能端到端学习，在多真实流中验证了不同项目属性。

链接：https://arxiv.org/abs/2607.09889

作者：Siddharth Pal, Viktoria Rojkova

英文摘要：Fixed-state sequence models compress an unbounded past into a bounded state, which caps their associative recall at roughly the state dimension; attention escapes the cap by keeping a key-value entry for every token, at quadratic compute and a cache that grows with the sequence. We study the middle ground: a sparse cache that allocates a slot only when an input is novel, so its size tracks the number of distinct items rather than the number of tokens. The allocation rule is the DP-means clustering rule, the small-variance limit of a Dirichlet-process mixture, used not as latent-variable inference but as the key-value memory operator for a deep recurrent backbone. We develop it in two forms, a static cache with a fixed concentration and a surprise-adaptive variant whose concentration follows the recent novelty rate. On a controlled associative-recall benchmark with redundancy we show that the cache matches full-attention recall while storing only the distinct items, that it dominates a fixed-budget eviction cache on the recall-versus-size frontier, and that on a state-space backbone it answers both a recall query and a long-range aggregate at the lowest memory of any model tested. The allocation is learnable end to end: a two-parameter novelty-threshold gate trained on the task loss alone recovers the rule exactly, whereas an over-parameterized gate fails, so the operative ingredient is the inductive bias rather than capacity. The evidence is a family of controlled mechanism studies at modest scale, with the distinct-items property confirmed on four real streams (recommendation, systems logs, clinical events, and insurance claims); a real-backbone, real-corpus language validation is pursued in a companion study.

AI 总结：研究大语言模型中MLP层以最优速率存储事实知识的现象，开发首个与Transformer兼容的事实存储MLP闭式构造，具有最优存储缩放等特性，所需参数更少，还能用于事实召回任务及实现模块化事实编辑。

链接：https://arxiv.org/abs/2607.10034

机构：Institute for Computational & Mathematical Engineering, Stanford University（斯坦福大学计算与数学工程研究所）； Department of Computer Science, Stanford University（斯坦福大学计算机科学系）； Department of Computer Science and Engineering, University at Buffalo（布法罗大学计算机科学与工程系）

作者：Roberto Garcia, Jerry Liu, Ronny Junkins, Sabri Eyuboglu, Atri Rudra, Christopher Ré

英文摘要：Large language models (LLMs) store factual knowledge in their parameters. While recent work has shown that this knowledge resides in MLP layers, existing constructive and mechanistic interpretability models of fact-storage in LLMs fail to explain the surprising empirical phenomenon that they store facts at an information-theoretically optimal rate. In this work, we develop a theoretical account of this phenomenon. We develop the first Transformer-compatible fact-storing MLP closed-form construction that satisfies the following three properties empirically observed in LLMs: it (i) attains optimal fact storage scaling, (ii) handles arbitrary input/output geometries, and (iii) works inside Transformers. Key to our work is to analyze the decoding margin of MLPs, whereas prior work only studies MLP fact storage. Under isotropic embeddings, our construction achieves information-theoretically optimal storage capacity scaling and requires $10$-$104\times$ fewer parameters at matched fact count than prior constructions. For arbitrary key and value embeddings, we show that our construction attains the same storage capacity scaling, up to penalization factors depending on the embedding geometries. Moreover, we demonstrate that our constructed MLPs can be used within Transformer blocks for factual recall tasks at optimal capacity scaling, requiring $15$-$63\times$ fewer parameters at matched fact count than prior constructions. Finally, as a proof-of-concept, we show that fact-storing MLPs enable modular fact editing by swapping a Transformer's MLP with a new one.

AI 总结：研究递归推理模型测试时扩展缺乏原则机制的问题，提出能量引导递归模型（ERM），利用霍普菲尔德能量定义选择器，与能量技术集成，在数独等问题上达最优解，优于其他模型，为有效推理提供途径。

链接：https://arxiv.org/abs/2607.10128

机构：Institute of Fundamental and Frontier Sciences, University of Electronic Science and Technology of China（电子科技大学基础与前沿研究院）； School of Physics, University of Electronic Science and Technology of China（电子科技大学物理学院）； Key Laboratory of Quantum Physics and Photonic Quantum Information, Ministry of Education, University of Electronic Science and Technology of China（电子科技大学量子物理与光子量子信息教育部重点实验室）； Non-classical Information Science Basic Discipline Research Center of Sichuan Province, University of Electronic Science and Technology of China（四川省非经典信息科学基础学科研究中心）

作者：Yifei Zhao, Ying Tang

英文摘要：Recursive reasoning models address structured problems by repeatedly updating latent states of small neural networks. However, their test-time scaling lacks a principled inference mechanism: increasing depth or stochastic breadth generates more trajectories without a clear criterion for selection, and existing methods predominantly rely on additional q-heads or heuristic voting. Here, we develop the Energy-guided Recursive Model (ERM), which introduces an intrinsic selection principle based on explicit Hopfield energies. ERM leverages Hopfield-type memories of valid local or global structures to define the selector over candidate trajectories. The resulting energy seamlessly integrates with energy-based techniques such as parallel tempering to enhance sampling efficiency and ranking. With $D=64$ recurrent steps and $K=128$ candidates, ERM reaches optimal solutions on Sudoku ($98.97\%$), Pencil Puzzle Bench (PPBench, $88.04\%$) and Maze ($99.30\%$), improving upon recent Probabilistic Tiny Recursive Model and Equilibrium Reasoners. These results suggest that incorporating explicit energy functions into recursive reasoning offers a principled path toward more effective inference.

AI 总结：研究在语言模型推理中从候选推理链选正确答案的问题，提出跨模型共识方法，将其视为语言模型陪审团，通过错误去相关机制工作，在多个基准测试中表现出色，能预先表征，有定律和下限。

链接：https://arxiv.org/abs/2607.10139

作者：Ning Liu

英文摘要：Selecting the correct answer from a pool of candidate reasoning chains is the engine of test-time scaling, yet the standard selectors each carry a cost: self-consistency inherits the errors of the single model it resamples, and trained reward models need labeled data and transfer poorly off-distribution. We study a third signal, free at inference time: cross-model consensus, the degree to which independently trained models, each solving the problem once, agree on a final answer. We treat the panel as an LLM-jury, in which the structure of agreement, not any model's score of another, is the verification signal. Across seven benchmarks it selects correct answers better than self-consistency and far better than a model scoring its own candidates: on competition math it closes the entire gap to an oracle selector, while self-scoring closes almost none. The mechanism is error decorrelation: independently trained models err differently, so their wrong answers scatter while the correct one accumulates agreement. We make this precise with a parameter-free law, derived in closed form, that predicts consensus accuracy from three measured panel statistics to a mean absolute error of $0.03$ and exposes the method's ceiling: a shared-error floor where models share a misconception, near zero on math but non-trivial on science. Against four trained verifiers spanning discriminative, outcome, and generative reward models, the free LLM-jury matches the strongest inside their math training domain and is the top selector outside it. Cross-model consensus is thus a verifier we can characterize in advance: a law that says when to trust it, and a floor that marks where it cannot.

AI 总结：研究伊辛模型微观自旋配置上训练的无监督自动编码器如何学习宏观变量，通过量化多尺度学习揭示两种动态模式，利用递归动态轨迹分析证明预测误差诱导流场，建立学习动态观点并提供解释基础。

链接：https://arxiv.org/abs/2607.10285

机构：Stuttgart Center for Simulation Science, Cluster of Excellence EXC 2075, University of Stuttgart（斯图加特大学卓越模拟科学中心，卓越集群EXC 2075）

作者：Max Weinmann, Miriam Klopotek

英文摘要：We study how unsupervised autoencoders trained on microscopic spin configurations from the Ising model learn macroscopic, theory-relevant variables underlying the data-generating process. Without embedding domain knowledge, we mimic a typical discovery setting: We quantify learning across multiple spatial (coarse-graining) scales and reveal two distinct dynamical regimes controlled by main hyperparameters (model depth, width, and learning rate) -- a magnetization-dominated regime and an energy-dominated regime characterized by trade-offs in their representation quality. The first regime is a transitory state exhibiting dynamical scaling and fluctuations that follow an ordering-to-scale; the second gradually shifts resolution towards smaller scales relevant for the energy representation. Deep models trained at moderate and fast rates become arrested before reaching these regimes. With a novel analysis of recursive-dynamic trajectories, we demonstrate that prediction errors induce flow fields that produce a common trajectory topology across all representation spaces. A dynamical viewpoint of learning is established in which intrinsic properties expose the effects of forced changes in representation during training. We utilize the intuition that learning operates as a process driven far from equilibrium by fluctuations from the training data and optimizer to provide an interpretive basis grounded in both the physical world and the machine models that represent it.

AI 总结：研究潜在世界模型中可预测性的控制理论，指出当前以预测误差为目标不可靠，重新定义目标为预测与真实计划成本的差异，证明规划器次优性受其限制，通过实验证实相关结论。

链接：https://arxiv.org/abs/2607.10362

机构：University of Science and Technology of China（中国科学技术大学）； The Hong Kong University of Science and Technology（香港科技大学）； Hong Kong Baptist University（香港浸会大学）

作者：Hanzhe You, Yonggang Zhang, Maohao Ran, Zhiqin Yang, Zhenyuan Zhang, Wei Xue, Jun Song, Xinmei Tian, Yike Guo

英文摘要：Latent world models are trained to predict future states in a learned representation and are then deployed inside a planner that selects actions by simulating them forward. Current practice adopts the prediction error, the single- or multi-step rollout loss on held-out data, as the training and model-selection objective, on the assumption that a lower prediction error yields better control. We show that this assumption is unreliable for a structural reason: a planner does not query the model on the training distribution but on the states that its candidate actions reach, which generally leave the data manifold, so an error averaged over the data cannot by itself govern control. We therefore reframe the objective as the discrepancy between the predicted and the true plan-cost at the plan the planner commits to, and prove that the planner's suboptimality is bounded by twice this discrepancy, whereas the data-averaged prediction error neither bounds nor tracks it. Under a linear-control premise the discrepancy separates into two terms. The first is a small on-manifold residual, on which the predicted and true dynamics agree and which a spectral tax prices through the non-normality of the latent transition operator. The second is an off-manifold divergence, on which an action carries the state off the manifold and the two dynamics diverge; this divergence is the binding term and is bounded by no data-averaged error. Synthetic operators confirm the pricing formulas, and latent model-predictive control experiments confirm the decoupling: across seeds, the single-step validation error is essentially uncorrelated with control success, whereas a fidelity score on the planner-reachable measure tracks it.

AI 总结：研究大语言模型智能体KV缓存内存瓶颈问题，提出无需训练的区域感知逐出策略MemDecay，为令牌分配特定区域优先级和衰减率，经实验验证该策略能有效管理缓存，确立语义提示结构对KV缓存管理的作用并明确其与关注重要性的结合方式。

链接：https://arxiv.org/abs/2607.10582

机构：Independent Researcher（独立研究者）

作者：Venkatesha Matam, Keon Kim

英文摘要：Large language model (LLM) agents accumulate heterogeneous context, including system instructions, plans, user turns, retrieved documents, tool outputs, and intermediate reasoning, whose key-value (KV) cache can become a major memory bottleneck. Existing eviction policies generally apply the same attention- or recency-based rule to every token, ignoring semantic structure already available to the agent orchestrator. We introduce MemDecay, a training-free, region-aware KV-cache eviction policy. MemDecay assigns tokens region-specific base priorities and decay rates, refreshes retention scores when tokens receive attention, and evicts the lowest-scoring pages under a fixed cache budget while allowing critical regions to be pinned. We also provide a procedure for calibrating decay rates from measured attention lifetimes. We evaluate MemDecay at approximately 450 and 1,700 token contexts using Qwen2.5-1.5B and 3B. Across all settings, attention lifetimes differ by an order of magnitude across regions: system-token half-lives range from 148 to 189 decoding steps, compared with 14 to 16 for scratchpad tokens. Pinning preserves system-region facts at full-cache accuracy in every setting, while no baseline preserves more than 13 of 24. Region-aware retention remains effective as context grows, whereas recency-based retention collapses. Accumulated-attention retention performs better on unpinned content, however, and ablations identify attention-score normalization as the main limitation of the current formulation. These results establish semantic prompt structure as a robust signal for KV-cache management while clarifying how it should be combined with attention-based importance.

AI 总结：研究Transformer中归一化策略选择问题，通过实验发现不同任务中静态与自适应归一化的表现差异，提出AutoNorm-S训练策略，通过门控冻结计划减轻优化不稳定性，在多基准测试中取得良好性能。

链接：https://arxiv.org/abs/2607.10593

机构：KIIT（卡林加工业技术学院）

作者：Piyush Kaushik Bhattacharyya, Divyanshu Rai, Swastik Singh, Kumar Aakash, Ayush Ranjan, Krutika Verma

英文摘要： Normalization is a critical component for stabilizing Transformer training, yet the choice between static strategies such as Layer Normalization (LN) and adaptive alternatives remains largely task-dependent. In this paper, we investigate a key optimization challenge in differentiable normalization gating. Our experiments show that, on relatively stationary vision tasks, the high gradient variance introduced by Gumbel-Softmax gating can hinder convergence of the routing mechanism, causing learned gates to underperform simple random selection. In contrast, on non-stationary language modeling and classification tasks, sustained gating diversity enables the model to learn more effective layer-wise normalization policies. Motivated by these observations, we propose AutoNorm-S (Stabilized), a training strategy that mitigates optimization instability through a gate-freezing schedule. AutoNorm-S achieves competitive or improved performance across multiple benchmarks, outperforming adaptive normalization baselines on NLP datasets, including PTB and SST-2, while remaining competitive on standard vision benchmarks. These results suggest that decoupling normalization selection from optimization noise provides a practical and principled approach for adaptive normalization in Transformer architectures.

AI 总结：针对多元时间序列分析难题，提出MSC-OT方法，结合多尺度卷积与最优传输注意力，由多尺度卷积增强、最优传输正则化及自适应融合策略构成，在多数据集实验中表现良好，验证了组件有效性与协同作用。

链接：https://arxiv.org/abs/2607.10740

机构：Institute of Collaborative Innovation（协作创新研究所）； University of Macau（澳门大学）

作者：HaoChong Fu, Jian Xu

英文摘要：The analysis of Multivariate Time Series (MTS) plays an important role in a lot of real-world practical applications, but it still remains some challenging problem about capturing multi-granularity structural patterns and suppressing noise appropriately. Multi-Scale Convolution with Optimal Transport Attention (MSC-OT) is proposed in this paper. MSC-OT is a useful architecture to optimize the attention mechanism. It combines multi-scale convolution with Sinkhorn optimal transport method based on inverted embedding. The inverted embedding approach embeds each variable as a token and allows the model to capture cross-variate relationships better. MSC-OT consists of two part: (1) Multi-Scale Convolution Enhancement, that applies multi-scale convolutions to attention score matrices based on inverted embedding, capturing local structural patterns in the variate-interaction space induced by compressed temporal representations; (2) Sinkhorn Optimal Transport Regularization, that formulates attention computation as an optimal transport problem and employs iterative matrix scaling to ensure balanced information flow across variates. Adaptive Fusion Strategy utilizes softmax-normalized learnable weights to dynamically combine base attention, convolution-enhanced, and OT-regularized scores. Experiments on widely-used datasets, including ETT, Electricity, Traffic, Solar-Energy, and Exchange-Rate, show that MSC-OT achieves well performance in both short-term and long-term forecasting tasks. Ablation experiments further validate the effectiveness of each proposed component and their synergistic contributions to improving prediction accuracy for multivariate time series forecasting.

AI 总结：研究在未观察到混杂因素时识别异质治疗效果，核心方法是利用代理模型中潜在混杂因素，通过构造压缩可观测算子及相关分析来恢复潜在治疗效果等，贡献是处理过完备代理系统并给出相关扰动界估计。

链接：https://arxiv.org/abs/2607.10926

机构：Dartmouth College（达特茅斯学院）； Broad Institute of MIT and Harvard（麻省理工学院和哈佛大学布罗德研究所）； Hofstra University（霍夫斯特拉大学）

作者：Hamza Virk, Bijan Mazaheri, Yihren Wu

英文摘要：Identifying heterogeneous treatment effects under unobserved confounding is central in observational causal inference. In proxy models with a discrete latent confounder, prior Synthetic Potential Outcomes (SPO) [Mazaheri-Squires-Uhler '25] recover the mixture of treatment effects through recursively constructed scalar moments. We show that this sequence is one projection of a more fundamental object. Under the same population factorization assumptions, there is an exact compressed observable operator: after projecting onto the shared proxy signal subspace, the difference of two treatment-arm quotient operators is similar to the diagonal matrix of latent treatment effects. Its eigenvalues are the latent effects; its lifted left eigenvectors, after anchor normalization, recover the target-proxy feature matrix and then the latent mixture proportions. Every scalar SPO moment is a bilinear functional of a power of this operator. The resulting estimator handles overcomplete proxy systems, replaces high-order scalar inversion with finite-dimensional spectral analysis, and admits high-probability first-order perturbation bounds for treatment effects, feature rows, and simplex-projected mixture weights.

AI 总结：针对联邦学习中恶意攻击和数据异构性问题，提出利用截断二次损失的新鲁棒聚合规则，减轻现有方法偏差，在非凸损失函数和异构数据下实现最优阶拜占庭鲁棒学习，实验证明该聚合器鲁棒性优于其他技术。

链接：https://arxiv.org/abs/2607.10970

机构：Ocean College, Zhejiang University（浙江大学海洋学院）； City University of Hong Kong（香港城市大学）； Aalto University（阿尔托大学）

作者：Zhi-Yong Wang, Hao Nan Sheng, Werner Stefan, Hing Cheung So, Linqi Song, Weitao Xu

英文摘要：Federated learning distributes data among $n$ clients, making it vulnerable to malicious attacks and data heterogeneity, which together pose challenges for robust learning. To tackle this issue, centered clipping and Huber aggregators have been exploited for Byzantine robustness. In this paper, we first demonstrate their equivalence via convex conjugate theory, and show that they can yield biased solutions in the presence of outliers, leading to failure under high data heterogeneity and a substantial fraction of outliers. Next, we propose a new robust aggregation rule that utilizes the truncated-quadratic (TQ) loss, effectively mitigating the biases of existing methods, such as centered clipping and Huber aggregators. We show that our aggregator achieves order-optimal Byzantine-robust learning under nonconvex loss functions and heterogeneous data, ultimately enhancing the reliability of federated learning systems. Additionally, we provide a robust deviation estimation strategy for TQ, demonstrating its effectiveness. Furthermore, we show that TQ maintains robustness even when only an estimate of the number of Byzantine clients is available. Finally, experimental results on MNIST, Fashion-MNIST, and CIFAR-10, indicate that our aggregator provides better robustness performance than the competing techniques.

AI 总结：研究语言模型预训练中的数据协同效应，利用开放权重语言模型的观测变化估计领域间协同效应，其框架提高预测精度，恢复稳定估计，经训练模型验证能正确预测性能排名。

链接：https://arxiv.org/abs/2607.11052

机构：MIT CSAIL（麻省理工学院计算机科学与人工智能实验室）； Microsoft Research（微软研究院）； Harvard University（哈佛大学）

作者：Kimia Hamidieh, Lester Mackey, David Alvarez-Melis

英文摘要：Machine learning progress is often attributed to scaling model size and dataset volume, yet the composition of data can be just as consequential. Empirical findings repeatedly show that combining datasets from different domains yields nontrivial interactions. For instance, adding code improves mathematical reasoning, while certain mixtures introduce interference that reduces model performance. We refer to these effects collectively as data synergy, where the contribution of multiple domains exceeds or falls short of the sum of their isolated contributions. In this work, we formalize and quantify data synergy in language model pretraining. Leveraging observational variation across open-weight LLMs with diverse pretraining mixtures, we estimate both direct domain-to-benchmark synergy (how one domain contributes to performance on another) and a second-order domain-domain synergy (capabilities that require co-occurrence of multiple domains). Our framework improves predictive accuracy over domain-agnostic scaling laws and recovers stable synergy estimates. We validate these estimates by training models on predicted optimal and predicted anti-optimal mixtures and confirm that our synergy estimates correctly predict performance rankings.

AI 总结：研究针对LLMs推理令牌阶段低并行性问题，提出FastTPS方法，含无重新加载KV缓存串联、优化“RoPE”注意力、高度融合MLP及细粒度流水线调度，显著缓解内存瓶颈，提升推理速度与内存带宽利用率。

链接：https://arxiv.org/abs/2607.11211

机构：Anonymous Institution（匿名机构）

作者：Wenzong Yang, Danyang Zhang, Kun Cao, Tejus Siddagangaiah, Rajeev Patwari, Zhanxing Pu, Siyin Kong, Zijiang Yang, Hao Zhu, Varun Sharma, Yue Gao, Tianping Li, Fan Yang, Jicheng Chen, Yushan Chen, Fennian Zhao, Aaron Ng, Elliott Delaye, Ashish Sirasao, Sudip Nag

英文摘要：The popularity of large language models (LLMs) escalates an ongoing demand for effective inference. However, due to the sequential processing of tokens during the token phase in decoder-only LLMs inference, the inherent low parallelism leads to reduced throughput and suboptimal utilization of the computing units on artificial intelligence (AI) accelerators, particularly when handling long-sequence inputs that impose significant memory overhead. Recently, many reported methods have been developed as potential solutions, since they emerge with numeric deviation. This paper presents FastTPS, a high performance and low-precision loss method for accelerating the token-phase in LLM inference on general AI accelerators which includes three key components: (1) AI accelerator-enabled reloading-free KV Cache concatenation which decreases memory access overhead as well as enables full fusion of Attention, (2) high-efficiency and high-accuracy 'RoPE' attention based on the tiling optimized FLAT, and (3) highly-fused MLP with fine-grain pipeline scheduling. Our results confirm that FastTPS significantly alleviates memory bottlenecks in the token phase, delivering a 6x speed improvement (compared to none-fusion) on an AMD Ryzen AI 300 series NPU with BF16 precision while sustaining 93% peak memory bandwidth utilization during Phi3-mini-4k-instruct inference.

AI 总结：研究刚性和冲击主导型偏微分方程求解问题，提出SPARC-Net架构及训练框架，通过自适应多尺度频谱编码器等解决相关问题，经多基准验证比普通PINNs有显著改进，如降低各方程误差，还进行多项分析。

链接：https://arxiv.org/abs/2607.11310

作者：Divyavardhan Singh, Dimple Sonone, Hammad Mohammad, Kishor Upla

英文摘要：Physics-Informed Neural Networks (PINNs) provide a meshless approach for solving partial differential equations (PDEs), but suffer severe degradation in stiff and shock-dominated problems, where small PDE residuals can correspond to globally inaccurate solutions. We show these failures are multi-causal, arising from the concurrent interplay of (i) spectral bias against sharp features, (ii) imbalanced multi-term optimization and loss-weight collapse, (iii) violation of temporal causality, and (iv) under-resolved collocation. We present SPARC-Net, a unified architecture and training framework that jointly addresses all four pathologies. SPARC-Net leverages an adaptive multi-scale spectral encoder with a learnable spectral gate, a gated residual backbone, adaptive activations, and a hard-constraint output ansatz that exactly enforces initial and boundary conditions, structurally eliminating loss-weight collapse. Training employs stabilized gradient-norm loss balancing, floored causality-respecting residual weighting, and residual-based adaptive collocation (RAD). Validated against exact analytic and high-order spectral reference solutions across four canonical benchmarks -- viscous Burgers', Allen-Cahn, convection (beta=30), and reaction -- SPARC-Net yields substantial improvements over vanilla PINNs: relative L2 error drops from 1.47e-1 to 1.14e-1 on Burgers' (22% reduction), 9.93e-1 to 5.78e-2 on Allen-Cahn (94% reduction), and 9.82e-1 to 3.54e-3 on reaction (100% reduction). A characteristic-coordinate encoder for hyperbolic transport further reduces convection error from 5.14e-1 to 9.88e-5 (100% reduction). We report five-seed mean +/- standard deviation errors, Wilcoxon significance tests, full ablation studies, hyperparameter sensitivities, an extension to the 2D heat equation, and comparisons against parameter-matched baselines.

AI 总结：针对观测表格数据因果发现难题，提出DAG-FM基础模型，通过自回归阶段、表格交互块及叶专家混合处理复杂情况，经迭代推理构建DAG，在合成及真实数据集上性能优于传统算法和其他模型。

链接：https://arxiv.org/abs/2607.11510

机构：Zhejiang University（浙江大学）； Tsinghua University（清华大学）

作者：Yikang Chen, Zhengkang Guan, Haoyuan Qian, Peng Cui, Yi Yang, Kun Kuang

英文摘要： Causal discovery from observational tabular data remains fundamentally challenging, primarily due to the heterogeneity of underlying causal mechanisms and the high-dimensional combinatorial search space of Directed Acyclic Graphs (DAGs). In this paper, we propose \textbf{DAG-FM}, a novel foundation model architecture that amortizes causal discovery. Unlike direct matrix prediction, DAG-FM decomposes the causal discovery process into two auto-regressive stages using two specialized Transformer-based sub-modules: a leaf-node predictor and a parent-node predictor. To effectively model complex row-column interactions, we adopt a robust tabular interaction block to output feature-wise representations. Crucially, to handle diverse and unknown Functional Causal Model (FCM) assumptions in real-world scenarios, we introduce Mixture-of-Leaf-Experts (MoLE), allowing the model to dynamically route and adapt to identifiable mechanism families. Through an iterative inference algorithm, DAG-FM seamlessly extracts causal orderings and constructs valid DAGs. Extensive experiments demonstrate that DAG-FM achieves state-of-the-art performance on both synthetic benchmarks and complex real-world datasets, significantly outperforming traditional classical algorithms and recent foundation models in both accuracy and scalability.

AI 总结：研究Transformer的理论理解，通过提出手工权重等分析其表达能力。针对其可学习性研究少的问题，受损失景观分析启发，为用Transformer学习C-RASP结构提出初步样本复杂度界限。

链接：https://arxiv.org/abs/2607.11760

机构：Mila & Université de Montréal（米拉与蒙特利尔大学）； University of Oxford（牛津大学）； Saarland University（萨尔兰大学）

作者：Michael Rizvi-Martel, Satwik Bhattamishra, Guillaume Rabusseau, Michael Hahn

英文摘要：A theoretical understanding of Transformers is crucial to better understand the capacities and limitations of large language models (LLMs). There is much work analyzing the expressivity of attention-based models. By proposing handcrafted weights or using computational complexity arguments, a large amount of past theoretical works have sought to characterize which tasks are and which are not in the hypothesis class of Transformer models. However, little work investigates the learnability of such solutions. In this work, we make progress towards this goal. Inspired by recent loss landscape analysis work, we propose preliminary sample complexity bounds for learning C-RASP constructions with Transformers.

AI 总结：研究多模态大语言模型安全机制权衡问题，提出输出感知安全护栏范式，通过在模型隐藏状态空间预测及多实例对比学习训练分类器，减少过度拒绝，匹配安全性能，保留模型效用与安全能力。

链接：https://arxiv.org/abs/2607.09697

机构：Lanzhou University（兰州大学）

作者：Jiayi Li, Kun Zhan

英文摘要：Existing safety mechanisms for multimodal large language models (MLLMs) face a fundamental trade-off between safety and utility. Model fine-tuning achieves robust safety but compromises general utility. Input-side safety guardrails offer a lightweight alternative, yet they suffer from severe over-refusal, indiscriminately blocking benign queries or those the model could have safely answered through refusal or advisory responses. We identify that the root cause of over-refusal lies in the input-aware paradigm: safety guardrails make safety decisions without considering whether the model itself is capable of generating safe responses. Usually, MLLMs already possess intrinsic safety mechanisms that can transform harmful inputs into harmless outputs, but input-side safety guardrails override this capability, degrading user experience. Motivated by this insight, we propose a paradigm shift toward output-aware safety guardrails. Our method operates within the model's hidden state space to predict whether the forthcoming generation will be unsafe before it is fully produced. By training a lightweight classifier via multi-instance contrastive learning on hidden state representations, our approach distinguishes between inputs that will lead to unsafe outputs and those that will not, even when the inputs themselves contain risky elements. This enables precise intervention only when the model's actual response would be harmful. Extensive experiments demonstrate that our output-aware safety guardrail matches the safety performance of existing methods while drastically reducing over-refusal, preserving the model's utility and built-in safety capabilities. Code is available at: https://github.com/kunzhan/OutGuard

AI 总结：本文围绕用于6G智能的联合嵌入预测架构（JEPA）展开。介绍其训练机制等，通过波束管理案例说明能提升标签效率与鲁棒性，还指出在多时间尺度预测等方面存在开放挑战。

链接：https://arxiv.org/abs/2607.09798

机构：IDCoM, University of Edinburgh（爱丁堡大学IDCoM研究所）； KTH Royal Institute of Technology（瑞典皇家理工学院）； SnT, University of Luxembourg（卢森堡大学SnT）； Department of Electronic Systems, Aalborg University（奥尔堡大学电子系统系）； School of Computer Science and Informatics, University of Liverpool（利物浦大学计算机科学与信息学院）； Division of Electronics and Electrical Engineering, Dongguk University（东国大学电子与电气工程系）； School of Engineering and Physical Sciences, Heriot-Watt University（赫瑞瓦特大学工程与物理科学学院）； Department of Electrical Engineering, Chalmers University of Technology（查尔姆斯理工大学电气工程系）； Department of Electrical and Computer Engineering, San Diego State University（圣地亚哥州立大学电气与计算机工程系）

作者：Sheikh Salman Hassan, Irshad A. Meer, Almoatssimbillah Saifaldawla, Yan Kyaw Tun, Mustafa Ozger, Madyan Alsenwi, Nguyen Van Huynh, Woong-Hee Lee, Cedomir Stefanovic, Mathini Sellathurai, Henk Wymeersch, Tharmalingam Ratnarajah

英文摘要： Sixth-generation (6G) networks are moving toward AI-native operation, where learning modules are embedded across the radio access network (RAN), edge, and core. This transition requires learning from limited labels, heterogeneous wireless and network data, partial observations, non-stationary propagation, and latency-constrained control loops. Joint-embedding predictive architecture (JEPA) is a promising self-supervised paradigm for this setting because it predicts missing or future representations in latent space instead of reconstructing raw measurements or using contrastive negative samples. This article presents a wireless-oriented tutorial on JEPA for 6G intelligence. We define the JEPA training mechanism, describe how CSI, beam measurements, KPIs, topology graphs, and sensing observations can be tokenized and masked, and position the learned encoder as a predictive representation layer for RAN, O-RAN, edge, and core functions, with task-specific heads or controllers producing final decisions. Then we present an illustrative, beam-management case study suggesting that a wireless-aware target, specifically an auxiliary future beam-energy target during self-supervised pretraining, can improve label efficiency and robustness across shifted deployment conditions relative to a supervised source domain. Finally, we outline open challenges in multi-timescale prediction, action-conditioned modeling, distributed training, trustworthiness, efficient deployment, benchmarking, and standardization.

AI 总结：研究针对网络安全新威胁无标注数据问题，提出SMETA-ZSL方法，通过对比微调、情景元学习和知识蒸馏等，从重叠语言描述学习语义原型并对齐行为特征，实现跨可见-未见类别的泛化，在7个基准测试中性能远超先前方法。

链接：https://arxiv.org/abs/2607.09936

机构：The University of Texas at El Paso（德克萨斯大学艾尔帕索分校）

作者：Ivan Alejandro Montoya Sanchez, Anantaa Kotal, Aritran Piplai

英文摘要：Cybersecurity systems must adapt rapidly to emerging threats. However, labeled data for new threat categories is unavailable when those threats first appear. Generalized zero-shot learning offers a natural solution by enabling recognition of unseen classes through auxiliary semantic knowledge rather than labeled examples. Large language models are particularly promising in this setting because they can convert unstructured CTI reports into semantic prototypes for emerging threats. However, applying language-driven zero-shot learning to cybersecurity is difficult due to strong semantic overlap between threat descriptions, heterogeneity between behavioral attributes and text, severe class imbalance, and open-set conditions where unseen threats are unknown during training. We propose SMETA-ZSL, that learns semantic prototypes from overlapping language descriptions through contrastive finetuning, aligns behavioral features through episodic meta-learning and knowledge distillation, and performs adaptive routing for generalization across seen-unseen classes. Across 7 benchmarks, SMETA-ZSL delivers the strongest overall generalized zero-shot performance under the strictest inductive setting, surpassing prior methods by 10.8 points on average, with gains up to 18.1 points. Github:https://github.com/Security-And-Intelligence-Lab-UTEP/SMETA-ZSL

AI 总结：研究非均匀随机图中距离保持嵌入，基于地标嵌入分析最短路径近似，扩展失真保证到全局平均并统一分析，引入GNN增强变体，使模型能从小图学习并推广到大型真实网络，保持或超越经典嵌入保真度。

链接：https://arxiv.org/abs/2607.10074

机构：Johns Hopkins University（约翰斯·霍普金斯大学）； Georgia Institute of Technology（佐治亚理工学院）

作者：My Le, Luana Ruiz, Souvik Dhara

英文摘要：Graph machine learning provides powerful tools for understanding complex networks and learning meaningful node representations. A central challenge, however, is designing embeddings with minimal distortion of both local and global functionals, such as shortest path lengths. Prior distortion guarantees for distance-preserving embeddings are worst-case in nature, producing overly pessimistic bounds that fail to capture the structure of typical large-scale networks. To address this, we analyze shortest-path approximation via landmark-based embeddings on inhomogeneous random graphs, a general model with type-dependent edge probabilities. By retaining shortest paths to a small set of reference nodes called landmarks, landmark-based methods effectively function as virtual graph spanners, where structural heterogeneity and controlled neighborhood expansion modeled via multi-type branching processes enable significantly tighter dimension-distortion trade-offs than classical worst-case bounds. We extend these guarantees to global, component-wide averages and unify the analysis across finite-type and continuous latent spaces through a novel metric sandwiching framework, establishing universal distortion bounds for general $L^2$ kernel models, including heavy-tailed and power-law networks. Finally, we introduce a GNN-augmented variant that replaces rigid, computationally expensive exact shortest-path queries with flexible, structure-aware neural surrogates. By leveraging the inherent alignment between graph neural message-passing and the dynamic programming principles of shortest-path algorithms, our approach demonstrates that models trained on small-scale random graphs learn to extract universal distance-preserving features, achieving robust generalization to large-scale, real-world networks that match or exceed the fidelity of classical, exact landmark-based embeddings.

AI 总结：研究CLIP风格双编码器对比学习中的模态差距，发现是InfoNCE公式在低温下的模式失败所致。提出xNCE方法，使用跨模态和模态内负对比对，能缩小模态差距，提升零样本分类性能且不牺牲判别几何。

链接：https://arxiv.org/abs/2607.10698

机构：Technical University of Denmark（丹麦技术大学）

作者：Fabian Mager, Hiba Nassar, Lars Kai Hansen

英文摘要：We study the modality gap in CLIP-style dual-encoder contrastive learning, where image and text embeddings remain misaligned despite being trained in a shared space. We argue that the gap is induced by a failure of the InfoNCE formulation with independent encoders. We conduct a uni-modal experiment with two independent encoders and identical initialization conditions and find that InfoNCE actively generates a gap at low temperatures. We provide a theoretical analysis of this phenomenon and show that the modality gap is indeed a mode-failure of InfoNCE, but only at low temperatures. We propose a simple modification called xNCE, which uses intermodal as well as intra-modality negative contrastive pairs. xNCE matches retrieval performance on MS-COCO while consistently reducing the gap even at low temperatures. Notably, xNCE improves zero-shot classification over the InfoNCE baseline across all benchmarks, whereas high-temperature InfoNCE and regularized InfoNCE both fail to do so, demonstrating that xNCE reduces the modality gap without sacrificing the discriminative geometry needed for transfer.

AI 总结：研究少样本多模态分类中轻量级头部校准不佳问题，核心方法是评估TabPFN作为即插即用分类头。贡献在于其在多数据集、多编码器和多模态评估中表现优异，能提升校准并保持准确率，为校准敏感多模态分类提供指导。

链接：https://arxiv.org/abs/2607.11007

作者：Jingxiang Zhang, Lujia Zhong, Zijie Zhu, Shuo Huang, Yuang Xu

英文摘要： Few-shot multimodal classification commonly attaches a lightweight head, such as $k$-nearest neighbors, logistic regression, or a linear SVM, to a frozen pretrained encoder. Although computationally efficient, these heads can produce poorly calibrated confidence scores, limiting their reliability in calibration-sensitive applications. We evaluate TabPFN as a plug-and-play, zero-gradient classification head for frozen image, text, and audio encoders. Across 22{,}820 evaluation episodes spanning 14 datasets, 11 encoders, and three modalities, TabPFN achieves the best mean rank among nine classification heads on both negative log-likelihood (NLL) and expected calibration error (ECE). At a representative setting, it reduces NLL by 48--62\% and ECE by 2.1--5.3$\times$ relative to the average of the eight baselines while matching or exceeding their average accuracy. Its accuracy advantage is conditional, concentrating at moderate-to-high shot counts and low-to-moderate feature dimensions ($k \ge 50$, $d \le 32$), and diminishing when labeled data are scarce, feature dimensions are high, or competing methods approach ceiling accuracy. In targeted backbone-adaptation experiments, replacing the trained linear head with TabPFN substantially improves calibration while preserving competitive accuracy. These results provide empirical guidance for using TabPFN as a training-free head in calibration-sensitive multimodal classification. To support transparency and reproducibility, we publicly release the source code, experiment configurations, and evaluation scripts in our GitHub repository: https://github.com/Jingxiang-Zhang/tabpfn-multimodal-embeddings.

AI 总结：研究旨在准确表示大气气溶胶群体，提出AeroMELD框架构建低维潜在变量以保留其数学结构，通过尺度-形状分解等实现，能准确重建多种分布及行为，为在潜在空间学习气溶胶过程演化奠定基础。

链接：https://arxiv.org/abs/2607.11073

机构：University of Illinois Urbana-Champaign（伊利诺伊大学厄巴纳-香槟分校）； Amazon（亚马逊）； Sandia National Laboratories（桑迪亚国家实验室）

作者：Ehsan Saleh, Saba Ghaffari, Wenhan Tang, Jeffrey H. Curtis, Lekha Patel, Peter A. Bosler, Nicole Riemer, Matthew West

英文摘要：Accurately representing atmospheric aerosol populations is essential for simulating aerosol-cloud interactions, radiative forcing, and ice nucleation, yet existing reduced schemes impose structural assumptions that limit their ability to capture composition diversity and mixing state. Machine-learning approaches offer more flexible representations, but standard autoencoders do not preserve the mathematical structure of aerosol populations and therefore cannot support physically meaningful process operators. We introduce AeroMELD (Aerosol Measure Embedding for Latent Dynamics), a mathematically grounded framework for constructing low-dimensional latent variables that retain this structure. We show that any permutation-invariant linear encoder must take a scale-shape decomposition, with total number concentration represented explicitly and latent shape given by a barycentric combination of per-particle embeddings. This aggregated latent state retains the diagnostic expressiveness of a Deep Sets model by moving the nonlinear post-aggregation stage into the learned diagnostic map while preserving latent linearity. Using particle-resolved data as ground truth, we encode weighted particle populations directly rather than binned aerosol states; size-resolved mass and number distributions serve only as diagnostic targets and visual summaries. The latent space accurately reconstructs these distributions, CCN spectra, optical coefficients, and immersion-freezing behavior while preserving the linear population structure needed for hybrid ML-physics models. Although the experiments focus on diagnostic reconstruction, the embedding is designed so that emissions and mixing can be represented exactly and nonlinear microphysical processes learned in a controlled latent space. This work establishes a foundation for learning aerosol-process evolution directly in latent space.

AI 总结：针对图欺诈检测中现有GNN检测器受节点属性不完整和类不平衡限制的问题，提出GFD-GC框架，通过模仿邻域结构进行组内聚合补全节点特征，引入置信度感知对比学习增强欺诈表示，实验验证该框架优于现有基线。

链接：https://arxiv.org/abs/2607.11107

机构：College of Computer and Information Science（计算机与信息科学学院）； Southwest University（西南大学）

作者：Junpeng Wu, Ye Yuan

英文摘要：Graph fraud detection plays a pivotal role in safeguarding the security and integrity of modern digital ecosystems. Graph Neural Networks (GNNs) are commonly adopted for graph fraud detection. However, the practical performance of existing GNN-based detectors is severely hindered by incomplete node attributes and extreme class imbalance within graphs. To mitigate these limitations, this paper proposes a novel framework for Graph Fraud Detection with Grouped attribute completion and Confidence-aware Contrastive learning, named GFD-GC. Specifically, it first imitates heterogeneous neighborhood structures to implement group-wise aggregation, which obtains informative complete node features by capturing fine-grained graph contextual patterns. Further, it introduces a confidence-aware supervised contrastive learning strategy to augment scarce labeled fraud nodes with high confidence pseudo-fraud nodes, which enhances the compactness of fraud representations and their separability from non-fraud nodes. Extensive experiments demonstrate the superiority of the proposed GFD-GC over state-of-the-art baselines on the graph fraud detection task, thereby providing an effective solution for real-world fraud scenarios.

AI 总结：研究在两个推理任务中具有学习初始化的端口哈密顿深度平衡模型，发现隐式计算存在问题，求解平衡常等于起始点，迭代次数与难度无关，模型未超两层MLP，追溯机制到梯度饥饿，提炼四测试诊断协议。

链接：https://arxiv.org/abs/2607.11116

机构：Independent Researcher（独立研究者）

作者：Joyjeet Singh

英文摘要： Deep equilibrium models promise input-adaptive implicit computation: harder problems should demand more solver iterations, and the solved equilibrium should encode the result of genuine iterative inference. We report a cautionary study of a port-Hamiltonian DEQ with a learned initialization on two reasoning tasks -- ProofWriter entailment over frozen DeBERTa embeddings and a BFS-verified graph-reachability benchmark -- in which the implicit computation is a silent no-op. Across tasks, seeds, and controlled ablation arms, the solved equilibrium equals the solver's start point to numerical precision, and bypassing the solver entirely changes test accuracy by +0.00 percentage points in 18 of 19 training runs. Controlled interventions falsify the tempting explanation: removing the anchoring term reproduces every result, and retraining with noise-decoupled starts yields a solver that converges to the noisy start while the decoder learns to ignore it. The single escaping run diverges instead ($\|h^{*}-z_0\|=171$), producing a co-adapted noise channel whose removal improves accuracy. Iteration counts are uncorrelated with ground-truth difficulty ($r=0.009$), and the full apparatus never outperforms a two-layer MLP on either task. We trace the mechanism to gradient starvation along two distinct routes, show that the standard zeroing ablation is confounded and gives wildly seed-dependent answers where the correct substitution test gives a stable zero, and distill a four-test diagnostic protocol for auditing claimed implicit computation. All experiments run on a single free Colab GPU; code, raw logs, and analysis scripts are released.

AI 总结：研究针对生成催化剂模型平板与体相连接缺失问题，提出对比表示学习模型CatRetriever，能准确检索体相候选物，还扩展其框架为吸附能目标体相发现管道，为连接催化剂生成模型与体相催化剂发现提供可扩展途径。

链接：https://arxiv.org/abs/2607.11712

作者：Jungho Oh, Woosung Kim, Dong Hyeon Mok, Jonggeol Na, Seoin Back

英文摘要：Inverse design is an emerging data-driven paradigm for efficiently navigating vast chemical spaces to discover new materials with targeted properties, and in the context of heterogeneous catalysis, surface generative models have recently advanced this goal by directly generating catalyst surface-adsorbate structures. However, these models typically operate at the slab level and do not provide the corresponding parent bulk structure, making it difficult to assess bulk-dependent properties such as formation energy, surface energy, crystallographic symmetry, and synthesizability. Here, we address this missing slab-to-bulk connection as a retrieval problem and introduce CatRetriever, a contrastive representation learning model that aligns slab and bulk crystal representations in a shared latent space. From a slab query, CatRetriever accurately retrieves plausible parent bulk candidates with R@1 > 91% and R@3 > 98% on both the in-distribution and holdout evaluation sets. We further extend the CatRetriever framework into an adsorption energy targeted bulk discovery pipeline that combines bulk retrieval, generative search space expansion, and adsorption energy distribution analysis. This workflow evaluates candidates by both structural compatibility with the query slab and their ability to access the target adsorption energy range across diverse surface environments. CatRetriever therefore provides a scalable route for connecting catalyst generative models with physically plausible and adsorption energy compatible bulk catalyst discovery.

AI 总结：研究旨在提升无训练语言模型推理能力，提出深度熵引导采样（DEGS）方法，利用逐层熵坍缩作为质量信号，结合序列似然性与深度熵结构，在多个模型和基准测试中达到无训练最优精度，在域外和难题测试中表现出色。

链接：https://arxiv.org/abs/2607.09693

机构：The Hong Kong University of Science and Technology（香港科技大学）

作者：Zibin Meng, Peng Xie, Kani Chen

英文摘要：Reinforcement learning (RL) has become the dominant paradigm for improving the reasoning capabilities of large language models, but it requires expensive training, curated data, and reward signals. Recent work shows that sampling from sharpened base-model distributions at test time recovers much of the RL gain, yet existing methods rely solely on output-layer likelihoods and ignore the transformer's internal forward-pass dynamics. We introduce Depth-Entropy Guided Sampling (DEGS), a training-free, test-time method that exploits layer-wise entropy collapse as an intrinsic quality signal. We observe that stronger reasoners -- including RL-posttrained variants -- exhibit a distinctive "late collapse": logit-lens decoded entropy stays elevated until deeper layers before converging. We define a per-sequence collapse depth $D(\mathbf{x})$ and a joint objective $π(\mathbf{x}) \propto p(\mathbf{x})^α\exp(βD(\mathbf{x}))$ that combines sequence likelihood with this depth-entropy structure, instantiated inside an MCMC power-sampling framework (DEGS-MCMC). Across three open-weight models and four reasoning benchmarks, this near-chance per-candidate signal compounds over the sampling trajectory into state-of-the-art training-free accuracy, with gains largest out of domain and on the harder splits -- exactly where likelihood alone falls short -- at single-digit-percent wall-clock overhead. DEGS narrowly trails an in-house GRPO reference on the math splits GRPO was trained for, yet surpasses it out of domain on GPQA for all three models, without any training, reward model, or labeled data.

AI 总结：研究大型自回归语言模型自我修正盲点问题，提出频谱代数理论SPARC，定义错误传播算子，推导激活阈值，证明基于强化学习的验证器-校正器训练收敛条件，实验验证定理，频谱预测与盲点率匹配。

链接：https://arxiv.org/abs/2607.09803

机构：European University of Tirana（地拉那欧洲大学）

作者：Ingrid Petrova, Luan Vejsiu

英文摘要：Large autoregressive language models exhibit a self-correction blind spot: they reliably fix identical errors when attributed to an external source yet fail to fix the same errors in their own outputs. Prior work has documented this phenomenon empirically, through controlled error injection, error-depth decompositions, RL-based verifier-corrector training, and intrinsic self-verification, but offers no formal model of why generating a token suppresses the ability to detect its error, no quantitative activation condition for correction markers, and no convergence guarantee for reinforcement-learning-based self-correction. We close these gaps with SPARC, a spectral-algebraic theory of self-correction in autoregressive generation. We define the error-propagation operator as the product of per-step attention Jacobians on the residual stream and prove that the blind spot arises if and only if the spectral radius of this operator is at least one. We derive a sharp activation threshold, given as a function of the spectral radius, that a correction marker must exceed, recovering the 89.3\% blind-spot reduction observed with a simple ``Wait'' marker. We further prove that RL-based verifier-corrector training converges at a rate proportional to the squared coupling strength over the square root of the number of samples if and only if the verifier-corrector coupling matrix has spectral norm below one, and that this criterion is invariant across residual-stream autoregressive modalities, unifying text LLMs and autoregressive image and video generation. Experiments across four backbones and a visual autoregressive probe validate every theorem, with spectral predictions matching measured blind-spot rates within 3.2\% RMSE.

AI 总结：研究四种训练后模式的Transformer语言模型，通过五个几何指标比较三种提示条件，发现多模态指令微调会使身份编码从方向重组为大小，此重组特定于该模式，还定位了W_1作为方法贡献。

链接：https://arxiv.org/abs/2607.09842

机构：Axis Dynamics SpA（轴动力公司）

作者：Jorge A. Castillo, Marco Torres Yévenes, Juan Carlos Lanas

英文摘要：We investigate whether identity-specifying system prompts produce statistically distinguishable geometric fingerprints in the hidden-state trajectories of four open-weight transformer language models spanning four post-training regimes: no training (Gemma-4-E4B base), multimodal RLHF (Gemma-4-E4B-it), RL distillation (DeepSeek-R1-Distill-Qwen-7B), and SFT (Qwen2.5-7B-Instruct). Three prompt conditions (an identity-specifying axis prompt, a length-matched generic-assistant prompt, and a 26-token vanilla baseline) are compared via five geometric metrics, principally the 1-Wasserstein distance between edge-wise distributions of Ollivier-Ricci curvature on k-NN trajectory graphs. Claims rest on trajectory-level permutation tests with multiple geometric controls (teacher-forced content controls, temporal-chain vs k-NN topology, ABT-projected k-NN, angular vs Euclidean graph construction, B=5000 permutations on borderline statistics). The central finding is a qualitative reorganization of identity encoding across the instruction-tuning boundary: in the base model the fingerprint is direction-coded (separation 0.034, p=0.002 under angular k-NN); in the multimodal instruction-tuned model it migrates into the magnitude (angular separation collapses to p=0.439 while Euclidean survives at p=0.042, and the mean norm of the first generated state inverts its length-ordering, being lowest for the identity prompt). This direction-to-magnitude reorganization is specific to the multimodal instruction-tuning regime, absent under RL distillation and SFT. A teacher-forced control attributes ~30% of the free-running cosine signal to prompt-driven effects. We position W_1 on edge-wise Ollivier-Ricci distributions on k-NN trajectory graphs as a methodological contribution of independent interest.

AI 总结：研究针对多专家示范下逆强化学习问题，采用狄利克雷过程先验的非参数贝叶斯方法，通过数据并行吉布斯采样联合推断潜在奖励类型及奖励，在ObjectWorld评估效果良好，还实现并行加速并刻画权衡。

链接：https://arxiv.org/abs/2607.09886

作者：Sai Anirudh Katupilla, Shreeya Dasa Lakshminath

英文摘要：Inverse Reinforcement Learning recovers reward functions from expert demonstrations, but standard formulations assume that all demonstrations come from a single expert. When demonstrations are pooled from multiple experts with distinct preferences, parametric methods recover an averaged reward that fits no individual expert well. We implement Nonparametric Bayesian Inverse Reinforcement Learning with a Dirichlet Process prior over reward functions, allowing the number of latent reward types to be inferred jointly with the rewards themselves. Inference uses a collapsed Gibbs sampler combining a Chinese Restaurant Process update for cluster assignments with a Metropolis-Hastings update for reward weights, and soft value iteration as the inner planning routine. We evaluate on a 10x10 ObjectWorld grid with two and three ground-truth reward types. The serial sampler recovers K=2 with Adjusted Rand Index of 1.000, substantially outperforming a Maximum Entropy IRL baseline (ARI=0.000). Extension to K=3 shows that the sampler correctly identifies the number of clusters in all runs; assignment ARI of 0.48-0.58 reflects behavioral overlap between expert types that persists across grid instantiations, revealing that reliable K=3 evaluation on ObjectWorld requires controlled object placement rather than random seeding. We further parallelize the sampler across CPU cores using Ray on HPC hardware, achieving a peak speedup of 4.79x at 8 workers, and characterize a throughput-versus-accuracy tradeoff arising from the consensus merge heuristic used during state aggregation. Code and a containerized environment are available at https://github.com/dasashreeya/np_bayes_irl.

AI 总结：研究大语言模型强化学习中PPO-Clip的探索崩溃问题，提出黎曼等距策略优化（RIPO）方法，通过保证黎曼流形上等距策略更新平衡探索与利用，在七个竞争级基准上显著超越现有算法。

链接：https://arxiv.org/abs/2607.10169

作者：Zhicheng Cai, Xinyuan Guo, Hanlin Wu, Mingxuan Wang, Wei-Ying Ma, Ya-Qin Zhang, Hao Zhou

英文摘要：Reinforcement learning (RL) has become a dominant paradigm for enhancing LLMs' reasoning capabilities. However, RL algorithms with PPO-Clip are inherently limited by exploration collapse. Subsequent works remain primarily heuristic and fail to identify the essential cause of PPO-Clip's failure. This work reveals the fundamental flaw of PPO-Clip: it implicitly measures policy discrepancy using Euclidean metric, which is theoretically inconsistent with the intrinsic geometry on the policy Riemannian manifold. This geometric mismatch results in overly conservative updates in low-probability regions while aggressive in high-probability regions, ultimately collapsing exploration. To correct this geometric flaw, we propose Riemannian Isometric Policy Optimization (RIPO), which guarantees isometric policy updates on the Riemannian manifold, effectively balancing exploration and exploitation. We further show that RIPO achieves a favorable bias-variance trade-off, which stabilizes optimization. Extensive experiments demonstrate that RIPO significantly surpasses existing LLM RL algorithms across seven competition-level benchmarks (up to 60% improvement over GRPO on AIME24).

AI 总结：研究针对偏微分方程求解器代码生成，引入RLVP强化学习训练后框架，通过混合验证器解决可验证性问题，在多PDE族训练单个策略，优于基线且有零样本改进转移，训练后小LLM表现良好，策略有组合性证据。

链接：https://arxiv.org/abs/2607.10474

机构：Massachusetts Institute of Technology（麻省理工学院）

作者：Pengfei Cai, Utkarsh Utkarsh, Alan Edelman, Christopher Vincent Rackauckas, Rafael Gomez-Bombarelli

英文摘要： Partial differential equations (PDEs) are foundational to modeling in science and engineering, but constructing reliable numerical solvers remains labor-intensive, demanding expert knowledge of discretization schemes, stability conditions, and boundary treatments. Recent work has begun to frame PDE solving as a code-generation task for large language models (LLMs), yet existing approaches operate primarily at inference time: relying on prompting, debugging, self-refinement, and test-time scaling rather than adapting the model itself. In parallel, reinforcement learning with verifiable rewards has emerged as a post-training paradigm for code and math reasoning, but its verifiers are typically binary: a compiler runs, or a test passes. Such signals discard the graded structure of scientific correctness, where two solvers may both execute and yet differ in solution accuracy by orders of magnitude. In this work, we introduce RLVP: Reinforcement Learning with Verifiable Physics, an RL post-training framework for multi-PDE solver code generation. RLVP addresses this verifiability gap with a hybrid verifier: hard program-validity checks ensure executability, while continuous physics rewards score function-space accuracy and PDE-residual consistency. A single policy is post-trained across diverse PDE families spanning hyperbolic, parabolic, elliptic, and incompressible-flow systems. RLVP improves over both pre-trained and supervised-only baselines on PDE benchmarks, and shows zero-shot improvement transfer to held-out PDEs. We show that a smaller LLM post-trained with RLVP can outperform prompting a frontier model on in-distribution PDE solver generation. The trained policy shows evidence of compositionality in numerical motifs: it recombines stencils, time-stepping schemes, and boundary-handling primitives learned from the PDEs used in training into generated solvers for unseen PDE problems.

AI 总结：研究大语言模型强化学习训练不稳定问题，提出ARMOR框架，通过锚展开利用离策略数据保留解决方案模式，混合优化重新制定策略目标实现可控探索，经实验验证可有效减轻验证崩溃，提升性能。

链接：https://arxiv.org/abs/2607.10481

机构：University of Science and Technology of China（中国科学技术大学）； Peking University（北京大学）； Dartmouth College（达特茅斯学院）

作者：Kexin Huang, Junkang Wu, Jinda Lu, Shuo Yang, Chiyu Ma, Jiancan Wu, Xiang Wang, Xiangnan He, Guoyin Wang, Jingren Zhou

英文摘要：Reinforcement learning (RL) has significantly enhanced the reasoning capabilities of large language models (LLMs), yet the training process remains notoriously fragile. In this work, we investigate a critical source of this instability: over-optimization, where models exploit training heuristics at the expense of generalizable reasoning. While reverse KL regularization is the standard defense against such degradation, our analysis reveals that it is often insufficient in this regime, as it fails to ensure comprehensive coverage of the reference distribution. To address this, we propose ARMOR (Anchor Rollout and Mixed Optimization for RL), a framework that shifts the paradigm from passive penalty to active sample stabilization. ARMOR comprises two key components: (1) Anchor Rollout, which leverages off-policy data from the reference policy to preserve established solution patterns; and (2) Mixed Optimization, which reformulates the policy objective to enable controlled exploration without relying on auxiliary losses. Extensive experiments on reasoning benchmarks validate that ARMOR effectively mitigates validation collapse, enabling sustained performance improvements over extended training horizons.

AI 总结：研究动态图上随机多臂强盗问题，基于滑动窗口混合识别结构条件，分析局部探索然后提交算法，建立次线性期望遗憾，还包括奖励感知策略及相关定理。

链接：https://arxiv.org/abs/2607.10571

机构：Department of Computer Science, University of Colorado, Boulder（科罗拉多大学博尔德分校计算机科学系）； INRIA, Paris（法国国家信息与自动化研究所巴黎分部）

作者：Sourav Chakraborty, Amit Kiran Rege, Claire Monteleoni, Lijun Chen

英文摘要：We study stochastic multi-armed bandits on dynamic graphs, where arms correspond to the vertices of a network with time-varying edges. In this setting, the learner is restricted to local movement, selecting only its current node or an immediate neighbor at each round. This constraint decouples best-arm identification from exploitation: even after the optimal arm is identified, the learner may remain unable to reach it through the evolving topology. We identify a process-agnostic structural condition, based on sliding-window mixing, that ensures the graph's intrinsic walk remains stable for both exploration and navigation. Under this regime, we analyze a family of local explore-then-commit algorithms and establish sublinear expected regret. Our framework includes a reward-aware strategy, for which we prove a worst-case safety theorem and a separate performance gain theorem.

AI 总结：研究资源受限设备上基础模型的最优持续微调问题，将其建模为约束马尔可夫决策过程，提出基于强化学习的演员评论家算法，实验表明该方法在准确率上优于同预算微调方法，且大幅减少微调步骤。

链接：https://arxiv.org/abs/2607.10694

机构：Athens University of Economics and Business（雅典经济与商业大学）

作者：Thomas Tsouparopoulos, Iordanis Koutsopoulos

英文摘要：We study the problem of optimal continual fine-tuning for a pre-trained Foundation Model deployed at a resource-limited device. At each time slot, a new batch of training data arrives, and the controller is faced with two options: either use the data to fine-tune the model and incur a compute cost, or do not fine-tune the model and discard the data. After the decision, the performance of the current model is measured in terms of an application-specific performance metric such as classification accuracy. Our objective is to learn an optimal policy that determines \emph{when to fine-tune the model} on a single task (e.g., sentiment analysis), under a finite compute budget. We formulate this online decision-making problem as a constrained Markov Decision Process, where the system state captures three essential aspects: (\textit{i}) model's performance, (\textit{ii}) computational budget, and (\textit{iii}) data distribution relevance to historic data encountered up to that point. The transition to the next state is stochastic and therefore, we propose a reinforcement learning-based method to solve this problem, namely the \emph{actor-critic} algorithm. We also consider the special case where the performance of fine-tuning for a given model can be predicted or estimated prior to decision; in this case the problem becomes a Dynamic Programming one. Experiments with a large pre-trained model on a widely-used text classification dataset demonstrate that our method consistently outperforms fine-tuning approaches with the same compute budget by more than $4\%$ in terms of accuracy and achieves $97\%$ of full-parameter fine-tuning accuracy while requiring only $25\%$ of the fine-tuning steps.

AI 总结：研究针对大语言模型训练范式易加剧搜索代理幻觉的问题，提出弃权感知强化学习（AWA-RL），利用模型能力动态塑造弃权奖励，并引入RA-F1指标，实验表明该方法有效提升了搜索代理的性能和可靠性。

链接：https://arxiv.org/abs/2607.10738

作者：Fengji Zhang, Tianyu Fan, Yuxiang Zheng, Xinyao Niu, Chengen Huang, Jacky Keung, Bei Chen

英文摘要：Recent advances in equipping Large Language Models (LLMs) with search tools and outcome-reward reinforcement learning (RL) have achieved new state-of-the-art results on open-domain QA tasks. However, we argue that current training paradigms harbor a critical vulnerability: they predominantly reward correct answers but fail to penalize fabricated ones when retrieval fails, thereby implicitly exacerbating hallucinations. To address this, we propose Abstention-Aware Reinforcement Learning (AWA-RL), which dynamically shapes the abstention reward utilizing the model's query-specific prior capabilities and continuous on-policy training observations. We also introduce a novel metric, RA-F1, to measure the capability-reliability trade-off. Compared to non-abstaining baselines, AWA-RL boosts absolute precision by up to 10.3% and overall RA-F1 by 2.9%, with only marginal sacrifice in raw accuracy. These results confirm that AWA-RL successfully yields highly capable and reliable search agents. The code, data, and model weights are publicly available at https://github.com/zfj1998/AWA-RL.

AI 总结：研究针对CT诊断阶段选择协议差异及现有方法局限，提出策略驱动的CT智能体（PD-CTAgent）。通过临床结构抽象模块和知识引导诊断控制模型，实现阶段感知诊断推理，实验验证其在多数据集上的有效性与临床一致性。

链接：https://arxiv.org/abs/2607.10748

机构：University of Science and Technology of China（中国科学技术大学）； Technical University of Munich（慕尼黑工业大学）

作者：Yanmeng Dong, Han Li, Yujia Li, Jingsong Liu, Xun Ma, Yanzhu Hu, Zhengyang Xu, Zhicheng Li, Nassir Navab, Shaohua Kevin Zhou

英文摘要：Computed Tomography (CT) diagnosis often relies on dynamic selection of imaging phases, such as non-contrast, arterial, or venous phases, based on preliminary findings, clinical suspicion, and diagnostic guidelines. This phase-wise decision process is critical for reducing unnecessary radiation exposure while supporting timely staging and treatment planning. However, phase-selection protocols can vary across hospitals, regions, and guidelines, while most existing CT-based AI methods assume that all phases are available and focus on static tasks under a fixed imaging phase, failing to model whether additional phases are required. This limitation stems from heterogeneous multi-phase representations, the need for knowledge-guided phase control beyond visual cues, and the lack of supervision for phase-sufficiency decisions in existing datasets. To address these challenges, we propose Policy-Driven CT-Agent (PD-CTAgent) for clinically consistent CT phase selection and diagnostic reasoning. PD-CTAgent introduces a Clinical Structure Abstraction Module (CSAM) to harmonize heterogeneous CT phases into a unified, phase-aware evidence representation. Based on this representation, a Knowledge-Guided Diagnostic Control Model (KDCM) evaluates phase sufficiency and iteratively requests additional phases when necessary. The policy-driven agent design further allows PD-CTAgent to flexibly follow different institutional, regional, or guideline-specific diagnostic protocols. Together, PD-CTAgent bridges static CT analysis and real-world clinical workflows. Experiments on two public datasets, LIDC and MCT-LTDiag, and one private dataset demonstrate its effectiveness and clinical consistency. Code will be made public upon acceptance.

AI 总结：研究大语言模型强化学习中PPO方向标准不足，提出预测性散度掩码，通过判断策略梯度步骤对散度的影响改进训练，针对离散softmax策略推导预测并开发轻量级估计器，提升不同模型规模和精度设置下的强化学习训练效果。

链接：https://arxiv.org/abs/2607.10848

机构：Tencent Hunyuan（腾讯混元）； UIUC（伊利诺伊大学厄巴纳 - 香槟分校）； NUS（新加坡国立大学）

作者：Xiangxin Zhou, Jiarui Yao, Penghui Qi, Bowen Ping, Jiaqi Tang, Haonan Wang, Tianyu Pang

英文摘要：Reinforcement learning for large language models (LLMs) typically relies on trust-region masks to stabilize off-policy updates. The dominant PPO-style approach uses the sampled-token importance ratio for two criteria: a proximity criterion, which asks whether the policy has moved too far from the behavior policy, and a direction criterion, which asks whether the update pushes it farther away. Recent work DPPO improves the proximity criterion by replacing PPO's ratio-based test with a probability divergence between the behavior and training policies. However, its direction criterion is still inherited from PPO. A token can be masked only when the sampled-token importance ratio moves away from one. We observe that this ratio-based direction criterion is a single-sample proxy that can disagree in sign with the change of the divergence that defines the proximity criterion. We therefore propose the predictive divergence mask, which asks whether the next policy-gradient step will increase or decrease the same divergence used by the trust region. For the discrete softmax policies used in LLM RL, we derive this prediction in closed form. Because production rollout engines expose only a truncated (top-K) view of the vocabulary, we develop two lightweight top-$K$ estimators for this prediction. Detailed analysis shows the divergence-based direction is better aligned with the realized change of the divergence than the sampled ratio, and the resulting masks improve RL training across model scales and precision settings.

AI 总结：研究在闭环去中心化交易所模拟器中，针对动态费用下的执行问题，采用构建最小闭环模拟器及小深度Q网络的方法，得出该方法能减少执行缺口，且优势集中在动态费用环境中的结论。

链接：https://arxiv.org/abs/2607.10960

作者：Wen-Ting Wang

英文摘要： Trader-facing dynamic fees are increasingly proposed for automated market makers (AMMs), but historical data do not identify how order flow would respond: trader-facing fees do not vary, trader types are latent, and a replayed tape is not a sequential decision environment. We therefore construct a minimal closed-loop simulator in which the missing signal exists by construction: two constant-product pools repriced by an equilibrium-inspired dynamic-fee rule, fee-sensitive noise flow, and closed-form CEX--AMM arbitrage. Equilibrium is used as a closure principle, not as an object the trader learns. Against a tuned benchmark ladder of schedule, planning, lookahead, and tabular policies, a small DQN is the only evaluated valid policy whose paired improvement over tuned one-step routing excludes zero. On a reserved final block of 1{,}000 seeds with completion forced to 1.0 for every policy, it reduces implementation shortfall under every tested intra-step ordering, by $13.3\bps$ of order notional under the pre-specified agent-last ordering, and the edge is concentrated in, and learned from, dynamic-fee environments: under constant fees the paired difference is indistinguishable from zero. The result is model-conditioned counterfactual evidence about execution control in AMMs, not evidence about historical traders, equilibrium play, or deployable profit.

AI 总结：研究基于人类专家成对轨迹比较的强化学习问题，提出受Bradley-Terry启发的理性模型捕捉不可比性并推断多维奖励函数，分析样本复杂度，评估模型在模拟环境中重建奖励函数及恢复帕累托前沿的能力与稳健性。

链接：https://arxiv.org/abs/2607.11432

作者：Simone Drago, Marco Mussi, Leonardo Bianconi, Alberto Maria Metelli

英文摘要：In this work, we study the reinforcement learning (RL) problem from pairwise trajectory comparisons provided by a human expert. We generalize preference-based RL by formalizing a novel setting in which the expert can also label trajectory pairs as incomparable, i.e., when neither trajectory dominates the other. We introduce the learning problem and the desiderata that its solution should satisfy. Then, we propose a novel Bradley-Terry-inspired rationality model that effectively captures incomparabilities and infers a multi-dimensional reward function, and we study its properties. We provide a sample complexity analysis for learning the model parameters when a dataset is available. Finally, we evaluate our model's ability to reconstruct a reward function that aligns with the expert's comparisons in simulated environments and to recover the Pareto frontier of policies, along with a robustness analysis across varying levels of expert rationality.

AI 总结：研究针对可验证奖励强化学习中推理路径反馈不足问题，提出SCOPE-RL框架，分两阶段优化，成功前添加奖励，成功后细化轨迹，并经评估协议验证，相比仅结果的GRPO提升准确率、减少推理令牌，且与其他方法互补。

链接：https://arxiv.org/abs/2607.11506

机构：Baidu Inc.（百度公司）； Shandong University（山东大学）

作者：Xiaojian Liu, Han Xu, Jianqiang Xia, Zhixuan Li, Ke Xu, Yiwei Dai, Xinran Chen, Changwo Wu, Yuchen Li

英文摘要：Reinforcement learning with verifiable rewards (RLVR) optimizes LLMs using sparse verifiable final-answer rewards. This sparse anchor reliably verifies whether a trajectory succeeds but provides no direct feedback on the reasoning path that produced it. Before success, prerequisite progress on hard problems receives no reward signal; after success, outcome rewards cannot distinguish well-organized correct trajectories from redundant or locally flawed ones. We introduce SCOPE-RL (Scaffolded Chain Optimization with Process Efficiency), a two-stage framework that densifies this anchor while retaining the GRPO update: Adaptive Scaffolded RL adds prefix-decomposed verifiable rewards on answer-hidden sub-question chains before success, and Quality-Aware Process RL applies correctness-gated process-shape rewards to refine correct trajectories after success. An expert-validated Step-Quality Evaluation Protocol evaluates useful-step density, error localization, and token efficiency beyond final-answer accuracy. On Qwen3-8B-Instruct trained on DAPO-Math and Big-Math, SCOPE-RL improves average accuracy by up to 11.2 pp and reduces reasoning tokens by up to 27.1% over outcome-only GRPO; the gains hold under GSPO and on Qwen3-0.6B-Instruct, indicating that reward-signal densification is complementary to policy-update-level RLVR advances. Code and data are available at https://github.com/tokencraft-lab/SCOPE-RL.

AI 总结：研究离线到在线强化学习中有限交互预算下的主动策略选择问题，提出基于未来性能上置信界主动选策略微调的方法，经实验验证其优于现有基线，能更有效利用预算，推动离线强化学习在现实系统的实际部署。

链接：https://arxiv.org/abs/2607.11720

机构：Virginia Commonwealth University（弗吉尼亚共同体大学）； University of Virginia（弗吉尼亚大学）

作者：Alper Kamil Bozkurt, Shangtong Zhang, Yuichi Motai

英文摘要：Background: Offline reinforcement learning (RL) enables effective policies to be trained from large, previously collected datasets and subsequently improved through limited online interaction. This offline-to-online RL (O2O-RL) paradigm is particularly promising in nonstationary domains where interaction is costly or potentially hazardous. Standard O2O-RL pipelines train multiple candidate policies offline, evaluate them using off-policy or online evaluation, and then deploy and fine-tune the policy with the highest estimated value. However, as in offline pretraining, fine-tuning performance is highly sensitive to the choice of algorithm and hyperparameters, making it risky to commit to a single policy. Objectives: We study active policy selection for fine-tuning under a limited interaction budget in O2O-RL settings. To our knowledge, this is the first work to address this problem. Methods: We formulate the problem by identifying a fundamental trade-off between allocating online interactions to policy evaluation, which helps identify high-performing policies, and allocating them to fine-tuning, which improves policy performance. We then propose an approach that balances this trade-off by actively selecting policies for fine-tuning based on upper-confidence bounds on their future performance. These bounds are derived from locally linear performance forecasts fitted to observations obtained through online evaluation. Results: Across a diverse range of experiments, the proposed approach consistently outperforms existing O2O-RL baselines. Conclusions: Actively selecting and fine-tuning policies uses limited online interaction budgets more effectively than either committing to a single policy or dividing the budget equally among all policies. Our framework also advances offline RL toward practical deployment in real-world systems where online interaction is costly or risky.

AI 总结：研究 PPVC 模块工厂灵活作业车间调度问题，通过对先进深度强化学习求解器进行三个扩展，得到时间滞后感知的学习策略，该策略在基准实例上表现出色，是无求解器调度器中最强的，还能快速重新规划。

链接：https://arxiv.org/abs/2607.11725

机构：Singapore Institute of Technology（新加坡科技学院）

作者：Ziheng Zhang, Wei Zhang

英文摘要：Prefabricated prefinished volumetric construction moves most building work into module factories, whose production floor operates as a flexible job shop. A major complication is decisive: long post-operation time-lags caused by concrete curing, watertightness ponding tests, and paint drying, during which a module is blocked while its workstation stays free. On benchmark instances grounded in an official national prefabrication guidebook, these lags inflate even the optimal reference makespan by about 67% on average, and ignoring them at decision time, then repairing to feasibility, is worse than every dispatching rule. We adapt a state-of-the-art dual-attention deep reinforcement learning solver through three minimally invasive, individually ablatable extensions: lag-aware dynamics with an admissible reward bound, two anticipatory lag feature channels, and liveness-masked operation- and station-type embeddings. With every extension disabled the implementation reproduces the original solver exactly, so all gains are attributable to the adaptations. We release a public, guidebook-grounded benchmark generator. On held-out instances the learned policy is the strongest solver-free scheduler: it reaches within about 4% of a constraint-programming reference and beats every dispatching rule and a genetic-algorithm metaheuristic, with its advantage widening under capacity contention, and a single size-mixed policy carries this lead across the trained range of factory sizes. It needs no solver, model, or license in the loop and re-plans within seconds of a disruption; where an exact solver can be deployed, that solver remains the quality ceiling, a boundary we map explicitly.

AI 总结：本文提出基于Transformer与人工蜂群算法结合的节俭神经架构搜索框架，通过动态熵机制防过早收敛，解决“冷启动”问题。在CIFAR-10及信用卡欺诈检测任务中表现出色，能发现高效、参数少的模型，适合边缘部署。

链接：https://arxiv.org/abs/2607.11826

作者：Romain Amigon

英文摘要：Neural Architecture Search (NAS) has automated the design of deep learning models but traditionally requires massive computational resources, often measured in thousands of GPU-days. In this paper, we propose a frugal and memetic NAS framework designed to democratize architecture design on consumer-grade hardware. Our approach combines the global macro-search capabilities of an autoregressive Transformer controller, trained via Reinforcement Learning (RL), with the local micro-exploitation of an Artificial Bee Colony (ABC) algorithm. To prevent premature convergence during the RL phase, we introduce a dynamic entropy mechanism that forces topological exploration upon detection of performance stagnation. Evaluated on a standard GPU (NVIDIA RTX 3060), our hybrid method effectively resolves the "cold-start" problem inherent in metaheuristics. By algorithmically penalizing network depth, our framework actively mitigates model bloat: on the CIFAR-10 dataset, it discovers an efficient architecture reaching 84.85% accuracy with only $\sim$174,000 parameters (significantly smaller than standard baselines like ResNet-20) in 3 hours of search time. Furthermore, we demonstrate the framework's flexibility by applying it to credit card fraud detection, directly optimizing the F1-Score on highly imbalanced tabular data to reach a F1-Score of 0.71 with a compact network of $\sim$4,600 parameters. These results suggest that our approach can yield tailored, accessible, and highly parameter-efficient deep learning models suitable for edge deployment.

AI 总结：研究为一类无记忆噪声过程的扩散模型开发基于GEXIT函数的守恒定律，统一离散和连续扩散似然性表征，利用局部性属性简化训练，通过在合成数据和标准基准上验证，揭示有限容量去噪器对不同噪声后验近似精度差异影响性能。

链接：https://arxiv.org/abs/2607.10067

作者：Ziv Aharoni, Henry D. Pfister

英文摘要：While autoregressive models optimize the exact data likelihood via the chain rule, diffusion models are typically trained with denoising objectives. We develop conservation laws based on generalized extrinsic information transfer (GEXIT) functions for a broad class of memoryless noise processes, showing that the data--model cross-entropy (CE) can be characterized exactly as an integral of local information-theoretic derivatives along the noise path. This yields a unified characterization of the likelihood for discrete and continuous diffusion, with the Gaussian case reducing to the well-known mutual information--minimum mean-square error (I-MMSE) relationship. An immediate implication is a locality property: one can compute the information-theoretic derivatives using only the marginal posteriors along the noise path. As a result, training reduces to learning the marginal posteriors by minimizing the negative log-likelihood. While the conservation law implies that the entropy does not depend on the noise path, finite-capacity denoisers approximate the posteriors with varying accuracy across noise types, leading to differences in performance. We validate these predictions on synthetic Markov sources and standard benchmarks, including text8 and CIFAR-10.

AI 总结：针对脑胶质瘤拉曼光谱数据集小且异质的问题，开发条件变分自编码器生成合成光谱增强真实训练数据，在严格协议下评估，结果表明该方法能提升分类性能，支持深度生成式增强可提高机器学习在相关应用中的鲁棒性。

链接：https://arxiv.org/abs/2607.10196

机构：National Institute of Research and Development for Biological Sciences（罗马尼亚生物科学研究与发展国家研究所）； University of Turku（图尔库大学）； Research Institute for Artificial Intelligence “Mihai Drăgănescu”, Romanian Academy（罗马尼亚科学院“米哈伊·德拉格内斯库”人工智能研究所）； University of Bucharest（布加勒斯特大学）

作者：Andrei Iuşan, Iulian Vasile, Daria Voiculescu, Ion Petre, Andrei Păun, Bogdan Oancea, Mihaela Păun

英文摘要： Access to sufficiently large biomedical datasets remains a major obstacle for machine learning in Raman spectroscopy-based diagnostics. In particular, for glioma analysis, datasets are typically small and heterogeneous, affected by acquisition-specific variability. This work investigates the utility of deep generative augmentation in such a small-cohort setting. We analyze glioma biopsy spectra acquired from 58 tumor samples and consider both binary IDH-status classification and 6-class methylation subtype classification problems. To address the limited size and imbalance of the dataset, we develop a conditional variational autoencoder ($β$-CVAE) capable of generating class-conditioned synthetic Raman spectra. The generated data are evaluated in Train-on-Synthetic, Test-on-Real (TS/TR) and Train-on-Synthetic+Real, Test-on-Real (TSR/TR) settings under a strict patient-isolated cross-validation protocol. Models trained exclusively on synthetic data underperform models trained on real spectra, indicating a substantial domain gap between synthetic and real distributions. However, augmenting the real training data with synthetic spectra consistently improves classification performance across multiple models. These findings indicate that, even with a limited number of independent patient samples, generative models can capture sufficient structure to provide useful regularization for downstream classifiers. We also investigate a reconstruction-based inference strategy, termed Classification by Reconstruction (CbR), in which class prediction is based on reconstruction error under different class conditions. Overall, the results support the use of deep generative augmentation as a practical strategy for improving machine learning robustness in Raman spectroscopy applications characterized by limited biomedical datasets.

AI 总结：研究针对大语言模型推理时激活控制问题，提出\textsc{Cobras}方法，将其视为残差流超球面上的薛定谔桥，从优化问题推导控制目标，使控制方向查询自适应，实验显示该方法性能优于基线且避免分布外性能下降。

链接：https://arxiv.org/abs/2607.10517

机构：The Pennsylvania State University（宾夕法尼亚州立大学）

作者：Seyed Arshan Dalili, Ajay Narayanan Sridhar, Vijaykrishnan Narayanan, Mehrdad Mahdavi

英文摘要：Activation steering offers a lightweight alternative to fine-tuning for controlling large language models at inference time. While many existing methods implicitly optimize a log-density-ratio objective between desired and undesired activation distributions, they do so heuristically rather than deriving it from a principled optimization problem. Moreover, these methods produce query-independent steering directions that can degrade performance on both in-distribution and out-of-distribution (OOD) inputs. We introduce \textsc{Cobras} (Conditional Optimal Bridge for Riemannian Activation Steering), which addresses both limitations by casting activation steering as a Schrödinger Bridge on the residual-stream hypersphere. This formulation yields, to our knowledge, the first principled derivation of the log-density-ratio steering objective from a well-posed optimization problem. Solving the bridge via entropic optimal transport and extracting the probability flow ODE recovers the widely used density-ratio gradient as a special case when the Sinkhorn potentials are uniform. Crucially, the Schrödinger potentials are evaluated at the current activation, making the resulting steering direction inherently query-adaptive. Empirically, across four models and three alignment axes (helpfulness, truthfulness, and detoxification), \textsc{Cobras} consistently outperforms prior activation steering baselines while avoiding the OOD degradation commonly observed in existing methods. The code can be found at https://github.com/arshandalili/cobras.

AI 总结：研究针对深度生成模型在风险敏感应用中尾部估计问题，提出历时样本整合（DSI）框架，通过整合随机训练轨迹各检查点样本，平均尾部波动，经理论形式化，实证显示其在固定预算下能大幅降低尾部估计误差，优于多种基线。

链接：https://arxiv.org/abs/2607.10810

机构：University of Melbourne（墨尔本大学）

作者：Shuning Zhao, Patrick Wong, Leran Zhang, Xiaolin Hu

英文摘要：Deep generative models are increasingly used as simulators for downstream decision-making under data scarcity, but in risk-sensitive applications their usefulness depends on rare adverse scenarios rather than typical samples. Standard generative objectives prioritize bulk distributional fidelity, leaving low-probability tails vulnerable to localized optimization noise and making tail-dependent functionals unstable under finite simulation budgets. We introduce Diachronic Sample Integration (DSI), a test-time inference framework that ensembles generated samples across checkpoints from a stochastic training trajectory. DSI targets a checkpoint-mixture distribution that averages checkpoint-specific tail fluctuations rather than relying on a single brittle endpoint. We formalize this mechanism through a finite-budget bias-variance theory. Empirically, across multivariate synthetic processes and high-frequency trading data, DSI substantially reduces tail-estimation error compared to single-checkpoint baselines under fixed simulation budgets, outperforming standard diffusion and state-of-the-art tail-aware baselines without modifying the generative objective.

AI 总结：研究高维中学习错误指定单指标模型的无限宽两层神经网络总体梯度流，通过摄动参数联合优化两层，证明隐藏链接函数相关分量能在预测时间尺度恢复，分析二次分量学习起始及权重经验测度奇异行为。

链接：https://arxiv.org/abs/2607.10869

机构：Unité de Mathématiques Pures et Appliquées (UMPA), ENS Lyon（纯数学与应用数学单位（UMPA），里昂高等师范学校）； CNRS（法国国家科学研究中心）

作者：Cédric Gerbelot, Jean-Christophe Mourrat

英文摘要：We study the population gradient flow of an infinitely wide two-layer neural network learning a misspecified single-index model in high dimension. The two layers are optimized jointly, with a perturbative parameter tuning the relative training speed between the first and second layer. This setting was considered by Berthier, Montanari and Zhou in \cite{berthier2024learning}, who conjectured a hierarchical learning scenario with explicit timescales as the second layer is trained faster than the first. In this paper, we prove that the constant and linear components of the hidden link function are indeed recovered within the predicted timescales, at sharp explicit thresholds. We then analyze the onset of learning of the quadratic component and show that the components learned at earlier stages continue to influence the dynamics in an essential way. Our proof is based on quantitative approximation results for singularly perturbed flows evolving near a manifold defined by integral constraints. At a phenomenological level, we also show that the empirical measure of the weights displays singular behaviour when reaching the quadratic component of the hidden link, with a small fraction of neurons growing significantly while the remaining ones rearrange to preserve the components already learned.

AI 总结：研究提出奇点空间这一用于信号表示的生成扩散框架，基于复平面奇点，利用基于变压器的扩散模型，具有可解释性、结构稳定性等特性，在一维伯格斯激波测试中表现良好，为瞬变主导信号提供实用基础。

链接：https://arxiv.org/abs/2607.10930

机构：Tel Aviv University（特拉维夫大学）

作者：Eli Bar-Yosef, Amir Averbuch, Eli Turkel

英文摘要：Generative models often represent signals as dense grids of amplitudes, blurring sharp transients that are crucial for the correctness of physical signals. We introduce Singularity Space, a generative framework that represents signals through complex-plane singularities, rooted in the classical pole-residue representation of meromorphic functions. We learn a latent space of physically constrained, per-signal singularity configurations to solve an inverse problem from degraded or partial observations. The framework has three key properties: interpretability, in which each generated singularity configuration corresponds to a set of physical parameters; structural stability, which mitigates Gibbs artifacts at discontinuities; and resolution-free output reconstruction on arbitrary grids without retraining or interpolation. Our framework employs a transformer-based diffusion model that directly predicts samples at complex-plane singularity coordinates, subject to geometric constraints during sampling. As a controlled test case for sharp-feature recovery, we evaluate our framework on 1D Burgers shocks, where each shock is represented by 32 predicted singularities (an $8\times$ reduction versus a 1024-point grid signal). Our framework preserves signal structure ($\text{TV ratio} \approx 1$) under unseen test-time observation noise, achieves a $4.2\times$ lower reconstruction error in zero-shot sub-resolution generalization than a grid-based baseline, and recovers physical parameters to $10^{-4}$ absolute error in-distribution. These results suggest that singularity-based representations may provide a practical foundation for other transient-dominated signals such as speech and biomedical signals, with potential extension to higher-dimensional domains.

AI 总结：研究引入粘性跳跃扩散（SJDs）统一掩码、连续和混合扩散。通过去噪危险匹配从单个去噪分类器估计相关量，SJD能恢复多种扩散类型，其反转解释各类型特征，还拓展了解粘核设计空间，在多数据集上表现优于恒等核混合方法。

链接：https://arxiv.org/abs/2607.10951

机构：Purdue University（普渡大学）

作者：Pascal Jutras-Dubé, Patrick Pynadath, Jeremy Lu, Yuan Gao, Ruqi Zhang

英文摘要：We introduce Sticky Jump Diffusions (SJDs), continuous-time Markov processes on $\mathbb R^d$ whose discrete anchors are token embeddings. In forward time, anchors release their mass at a hazard rate and the released mass diffuses in the continuous ambient space; time reversal couples a score-driven SDE with a sticky jump kernel whose rate and destination are fixed by flux balance with the forward law. We estimate the score and the per-anchor reverse hazards from a single denoising classifier via Denoising Hazard Matching, the hazard analogue of denoising score matching, with simulation-free cross-entropy training. SJD recovers masked diffusion, continuous diffusion, and hybrid diffusion as limits. Its reversal explains features that each family treats as given: the mask of masked diffusion carries no evidence about the source token because the unsticking kernel of every anchor collapses to the same absorbing point; the terminal projection of continuous diffusion is required due to the absence of atoms in its forward marginal, without which flux balance yields no reverse jumps; and the update rules of hybrid diffusion (commit rate, destination, and drift) all follow from flux balance rather than from separate design. Beyond these limits, the unsticking kernel becomes a design space: a cross-position blending corrupts each position toward a blend of its neighbors' clean values or embeddings, turning dependency structure such as spatial locality or a constraint graph into an inductive bias of the corruption itself, and improves over the identity-kernel hybrid on CIFAR-10, Text8, and Sudoku.

AI 总结：研究从稀疏有噪观测中发现积分 - 微分方程的记忆和非局部核这一逆问题，提出基于可微求解器框架，用约束柯尔莫哥洛夫 - 阿诺德网络参数化并施加物理约束，经训练和符号回归得到表示，实验表明构造约束比软惩罚更具鲁棒性。

链接：https://arxiv.org/abs/2607.11110

机构：University of Utah（犹他大学）

作者：Aruzhan Tleubek, Salah A Faroughi

英文摘要：Discovering the memory or nonlocal kernel governing an integro-differential equation (IDE) from sparse and noisy observations is an ill-posed inverse problem. Existing identification methods often rely on problem-specific analytical derivations, specialized observation requirements, or restrictive assumptions about the kernel, limiting their applicability across different classes of IDEs. In this work, we propose a differentiable-solver-based framework for discovering memory and nonlocal kernels directly from spatiotemporal observations. Within the solver, the unknown kernel is represented using a constrained Kolmogorov--Arnold Network (KAN) parameterization, with the physical constraints imposed through two different approaches: a Bernstein-polynomial-based Monotone--Convex KAN (MC-KAN), whose coefficient constraints enforce positivity, monotonic decrease, and convexity by construction, and a Chebyshev-based KAN (Cheb-KAN), in which the same properties are encouraged through soft penalty terms. After training, symbolic regression is applied to the learned kernels to obtain interpretable closed-form representations. We evaluate both methods on benchmarks spanning a one-dimensional Volterra equation, a one-dimensional viscoelastic wave partial integro-differential equation, and a two-dimensional nonlocal reaction-diffusion equation with an anisotropic coupled kernel. For the 1D problems, both methods recover the correct kernel functional form and achieve comparable solution-reconstruction accuracy. In contrast, for the sparse and noisy 2D nonlocal problem, the hard-constrained MC-KAN consistently achieves lower kernel reconstruction errors than the soft-constrained Cheb-KAN. Our results demonstrate that enforcing physically motivated shape constraints by construction provides greater robustness than soft penalties for multidimensional kernel discovery from sparse and noisy observations.

AI 总结：多模态AI中为文本叙事生成电影音效困难，BackgroundMellow框架将其视为编排与信号处理问题，通过主 - 专家架构、Tango2模型、背景音乐检索器及NLP模块实现，经实验验证在时间同步等方面有效。

链接：https://arxiv.org/abs/2607.11364

机构：Indian Institute of Technology Kharagpur（印度理工学院卡拉格布尔分校）

作者：Ajitesh Jamulkar, Aritra Hazra

英文摘要：Generating immersive, synchronized and cinematic audio for long-form textual narratives remains a significant challenge in multi-modal AI. While current Text-to-Audio (TTA) frameworks successfully synthesize isolated sound effects, they struggle with narrative cohesion, temporal alignment, and cinematic emotional depth. We present BackgroundMellow, a framework that treats story-to-audio generation as a precise orchestration and signal processing problem. This framework is enabled without ground-truth through a master-specialist agent architecture that decomposes text into precise and multi-layered audio cues, generates each category of sounds with suitable specialist model, and superimposes the soundscapes to create a unified and aligned audio segment. Our pipeline is built over Tango2 latent diffusion model for environmental synthesis alongside a novel Cinematic BGM Retriever mined from professional soundtracks. To automate the sound mixing process, we use an NLP based module that predicts precise audio parameters, like start time, duration, and relative loudness, based on the narrative timeline. We further empirically evaluate and show the efficacy of the proposed framework leveraging nearest-neighbor retrieval against a curated dataset of YouTube cinematic trailers to measure temporal synchronization, coverage, and spectral richness.

AI 总结：研究能否用训练好的生成模型更快生成多用户TR 38.901信道且保持空间相关性，提出物理感知条件SetGAN，经训练可分离大尺度与小尺度衰落相关信息，在UMa/NLoS基准测试中大幅加速信道生成且保持空间一致性。

链接：https://arxiv.org/abs/2607.11429

机构：Institute of Telecommunications and Multimedia Applications (iTEAM), Universitat Polit\`ecnica de Val\`encia (UPV), Spain（电信与多媒体应用研究所）； Beihang Valencia Polytechnic Institute (BVPI), China（北京航空航天大学瓦大理工大学）； Huawei Technologies, France（华为技术）

作者：Mauro Gonzalo Tarazona-Levano, David Lopez-Perez, Nicola Piovesan, David Gomez-Barquero

英文摘要：TR 38.901-based channel models such as Sionna are reliable, but generating many multi-user channel realizations remains expensive. This paper asks a practical question: can a trained generative model produce multi-user TR 38.901 channels faster than Sionna without losing the spatial correlations imposed by user geometry? To answer this question, we propose a physics-aware, geometry-conditioned SetGAN trained on Sionna reference data. The method separates large-scale received power from normalized small-scale fading, compresses the latter with principal component analysis, and learns the conditional channel distribution in a latent space while preserving geometry-dependent correlations. On the UMa/NLoS benchmark, the model keeps the received-power distributions close to the reference, with about 0.41 dB Wasserstein distance, and reproduces spatial-consistency profiles with mean deviations below 0.03 on median curves versus distance. In addition, it reduces elapsed generation time by a factor of 3.45 and CPU-total cost by a factor of 6.15 relative to Sionna under matched user positions in the fixed-position CPU-vs-CPU benchmark. These results show that a trained generative model can substantially accelerate TR 38.901 channel generation without breaking the spatial consistency needed to evaluate multi-user systems.

AI 总结：研究提出速度调度流匹配（VSFM）方法，放宽流匹配中样本恒速移动的隐含选择。通过特定多项式轮廓，预训练模型推理时无需重新训练和额外计算，FID降低，从零开始训练也有进一步提升，增益源于欧拉积分器局部截断误差。

链接：https://arxiv.org/abs/2607.11442

机构：Cherkasy State Technological University（切尔卡瑟国立技术大学）

作者：Vitalii Bondar

英文摘要：Flow matching trains a neural network to regress the conditional velocity along a linear interpolant between noise and data, and the number of network evaluations~(NFE) sets the cost of sampling. The straight-line interpolant carries an implicit choice: the sample moves at constant speed throughout the trajectory. We relax this choice and introduce Velocity Scheduled Flow Matching~(VSFM), which replaces the conditional target $x_1 - x_0$ with $v(t)(x_1 - x_0)$ for any nonnegative profile $v:[0,1]\to\mathbb{R}_{\geq 0}$ satisfying $\int_0^1 v\,dt = 1$. We study six polynomial profiles drawn from motion planning. The first use of VSFM is at inference time: a pretrained linear flow-matching model can be sampled under any admissible profile by integrating its ODE on a non-uniform $τ$-schedule, with no retraining and no additional computation; on CIFAR-10 this lowers FID by up to $19.8\%$. Training from scratch under a braking profile gives a further reduction of $17.4\%$ at $4$~NFE. Both gains follow from the local truncation error of the Euler integrator on the induced grid.

AI 总结：研究主动流动控制中控制器设计难题，提出用编码智能体直接搜索显式可执行反馈律的启发式学习范式及受限协议，经多基准评估，其启发式控制器性能优、可解释，是传统强化学习的可靠互补替代方案。

链接：https://arxiv.org/abs/2607.11565

机构：Mines Paris - PSL University Centre for Material Forming (CEMEF) CNRS（巴黎矿业 - 巴黎文理研究大学材料成型中心（CEMEF）法国国家科学研究中心）

作者：Paul Garnier, Jonathan Viquerat, Elie Hachem

英文摘要： Active flow control involves nonlinear dynamics, partial observations, and computationally expensive simulations, making controller design particularly challenging. Deep reinforcement learning (DRL) has emerged as a powerful framework for such problems, but its success typically relies on large numbers of simulator interactions and produces neural-network policies whose decision process often remains difficult to interpret. In this work, we investigate a different paradigm: instead of optimizing neural-network parameters, we use modern coding agents to search directly for explicit executable feedback laws. We introduce a constrained heuristic-learning protocol in which an agent iteratively proposes, evaluates, and revises controller implementations while interacting exclusively through the public benchmark interface. The proposed framework is evaluated on 13 active flow-control benchmarks spanning one, two, and three-dimensional problems and compared against the strongest available DRL baselines under identical simulation budgets. The discovered heuristic controllers match or outperform the best DRL policy in 10 of the 13 environments while remaining compact, interpretable, and directly inspectable. Beyond aggregate performance, the resulting controllers reveal physically meaningful feedback mechanisms, transfer successfully across more challenging configurations, and remain competitive under varying Reynolds and Rayleigh numbers, actuator counts, and observation sparsity. These results suggest that heuristic learning through coding agents constitutes a credible and complementary alternative to conventional reinforcement learning, combining competitive performance with physically interpretable controller representations. Prompts and source code are available at https://github.com/DonsetPG/fluid-heuristic-learning.

AI 总结：研究针对基于脑电的癫痫检测面临的注释稀缺和类别不平衡问题，提出DiffEEG自监督基础模型，通过去噪扩散预训练和强化学习微调学习通用神经表示，在癫痫检测中取得较好效果，证明该方法能以最少标记数据实现临床可部署监测。

链接：https://arxiv.org/abs/2607.11578

机构：Lebanese American University（黎巴嫩美国大学）； Institute of Applied Artificial Intelligence, TÉLUQ University（应用人工智能研究所，泰鲁克大学）

作者：Abdulkader Helwan, Lina Abou-Abbas, Hussein El Amouri, Belkacem Chikhaoui, Khadidja Henni

英文摘要：Deep learning for EEG-based seizure detection faces critical challenges: severe annotation scarcity and extreme class imbalance, where ictal events comprise less than 10\% of clinical recordings. We present DiffEEG, a 9.6M-parameter self-supervised foundation model that addresses both limitations through denoising diffusion pre-training and reinforcement learning (RL)-based fine-tuning. Pre-trained on 1.3M unlabeled segments from the Temple University Hospital Seizure Corpus (TUHSZ), DiffEEG learns generic neural representations via a 1D U-Net with multi-head self-attention. For downstream adaptation, a reinforced decision layer employs policy gradient optimization to directly maximize F1-score, prioritizing sensitivity to rare seizure events over overall accuracy. Under strict patient-wise evaluation (279 patients, Leave-One-Fold-Out), DiffEEG achieves 61\% accuracy and 59\% F1 for 4-class seizure subtyping, and 81\% accuracy with 85\% weighted F1 for binary detection, maintaining clinically viable seizure recall (59\%) despite extreme imbalance (6.7\% prevalence). Segment-level evaluation establishes an upper bound of 97.6\% accuracy, confirming strong architectural capacity. DiffEEG demonstrates that diffusion-based pre-training combined with metric-aware reinforcement learning enables clinically deployable seizure monitoring with minimal labeled data requirements.

AI 总结：研究针对离散扩散模型中专家组合方法的局限性，提出因子级组合框架FactorDiff，将样本分解为更小因子，通过动态路由使因子与相关专家匹配，在ARC - AGI基准测试中表现优于全局标量加权方案。

链接：https://arxiv.org/abs/2607.11758

机构：University of Toronto（多伦多大学）； Vector Institute for Artificial Intelligence（向量人工智能研究所）； Mechanical & Industrial Engineering, University of Toronto（多伦多大学机械与工业工程系）； Canadian Institute for Advanced Research (CIFAR)（加拿大高级研究所）

作者：Haozhe Huang, Yudong Xu, Abhijoy Mandal, Alán Aspuru-Guzik

英文摘要：Discrete diffusion models offer a powerful framework for solving complex reasoning tasks, particularly through compositional generation, which combines multiple pre-trained experts to generalize beyond their individual training data. Recent theoretical corrections introduce time-dependent mixing weights to better align composed diffusion dynamics with the intended target. However, these methods are fundamentally limited by working on a per-sample basis, treating each generated state monolithically and ignoring the potential spatial or functional specializations of different experts. In this work, we address this limitation by proposing FactorDiff - a factor-wise composition framework for diffusion models. We posit that samples can be further decomposed into smaller factors, and propose a sampling process that dynamically routes each factor to the most relevant expert. We instantiate this framework with spatial/pixel-level compositions and validate it on the ARC-AGI benchmark, demonstrating that simple factor-specific routing consistently outperforms complex global scalar weighting schemes on tasks that require logical consistency and spatial disentanglement.

AI 总结：研究以宏观经济预测为压力测试领域，评估五个模型家族在23个国家的表现。结果显示无模型能始终有强大预测性能，约束少的模型如ARIMA和NODE优于约束多的。指出结构先验不匹配数据生成过程时会起反作用，从业者应先测试结构是否有益。

链接：https://arxiv.org/abs/2607.09684

机构：GEMS Modern Academy（GEMS现代学院）； Vizuara AI Labs（Vizuara人工智能实验室）

作者：Vrishank Sai Anand, Prathamesh Dinesh Joshi, Raj Abhijit Dandekar, Rajat Dandekar, Sreedath Panat

英文摘要：Scientific Machine Learning (SciML) methods such as Neural Ordinary Differential Equations (NODEs), Physics-Informed Neural Networks (PINNs), and Universal Differential Equations (UDEs) are most effective when structural priors reflect reliable governing dynamics. We ask what happens when this assumption is violated. Using macroeconomic forecasting as a stress-test domain, we evaluate five model families, ARIMA, LSTM, NODE, PINN, and UDE, across 23 countries using sparse annual data, multiple temporal splits, and five random seeds. Our results show that none of the evaluated models achieve consistently strong forecasting performance, highlighting the difficulty of low-frequency macroeconomic prediction. However, a clear relative hierarchy emerges: less-constrained models, particularly ARIMA and NODE, consistently outperform more-constrained heuristic-prior models such as PINN and UDE. Rather than treating this as a rejection of SciML, we interpret it as a diagnostic result: structural priors can act as misregularizers when they do not match the data-generating process. We identify failure modes including prior misalignment, regime shifts, structural breaks, and optimization instability, and argue that SciML practitioners should test whether structure helps before assuming that more structure is beneficial.

AI 总结：针对低自相关二进制序列问题，提出结合汤普森采样与并行自回避游走的混合搜索框架，经GPU加速等优化，通过两阶段策略改进搜索。实验表明其在多序列长度上改进结果，获新最长序列，有效划分优先级，提供高性能优值因子最大化策略。

链接：https://arxiv.org/abs/2607.09688

作者：Blaž Pšeničnik, Borko Bošković, Jan Popić, Janez Brest

英文摘要：Low autocorrelation binary sequences problem (LABS) is a hard combinatorial optimization challenge with important applications in communications, signal processing, and satellite navigation. This paper proposes a hybrid search framework that combines Thompson sampling with parallel self-avoiding walks to adaptively allocate computational effort across restriction classes of the LABS search space. By modeling partitions as arms in a multi-armed bandit setting, the proposed method dynamically shifts search resources toward partitions that empirically produce higher merit factors while maintaining exploration of less-sampled regions. The approach is further accelerated through GPU-parallel execution, shared posterior updates, efficient neighborhood evaluation, and a Bloom filter for cycle prevention. In addition, we use a two-stage optimization strategy that first searches constrained partitioned skew-symmetric spaces and then refines the best candidates in the unrestricted space. Experiments on long binary sequences show that the proposed method improves the previously best-known results for 35 sequence lengths in the range $450 \le L \le 527$ and for $L=573$. In particular, we report a new longest sequence with merit factor exceeding $8.0$, obtained for $L=451$. The results also show that Thompson sampling effectively prioritizes partitions with better observed performance, confirming the value of online, data-driven resource allocation in LABS optimization. Overall, the proposed framework provides a scalable and effective strategy for high-performance merit factor maximization.

AI 总结：研究针对DPO性能依赖偏好数据质量问题，提出双层优化框架、无任务元知识驱动方法及结合中心差分近似与LoRA微调的可扩展训练方案，经实验验证该方法能在不同噪声率下提升训练性能。

链接：https://arxiv.org/abs/2607.09796

机构：Xi’an Jiaotong University（西安交通大学）

作者：Hua Qu, Yifan Li, Xiaodong Yuan

英文摘要：Direct Preference Optimization (DPO) has become an important method for aligning large language models (LLMs) with human preferences because it removes the need for explicit reward modeling and reinforcement learning optimization. However, its performance depends heavily on the quality of preference data, and noisy preference data in real-world settings can weaken alignment performance. To address this issue, we propose a bilevel optimization framework and prove, under certain assumptions, that this framework can recover the DPO optimum under clean data. We further derive a prior form for the learnable weighting function under asymmetric label-flipping noise. Considering that high-quality metadata may be difficult to obtain, we propose a task-agnostic meta-knowledge-driven method that enables meta-learning even when metadata is completely unavailable. To reduce the high cost of higher-order gradients in LLM meta-learning, we combine central-difference approximation with LoRA fine-tuning and develop a scalable training scheme. Experiments on TL;DR summarization and Anthropic HH single-turn dialogue show that the proposed method improves training performance over multiple DPO baselines under different noise rates.

AI 总结：研究类别不平衡问题，提出RUBRIC框架，将合成样本选择设为质量优化问题，用现实-效用权衡排序，能收紧泛化边界，经实验验证其在信用卡欺诈检测等任务中可提升F1宏和召回率，保持ROC-AUC并可进行lambda敏感性分析。

链接：https://arxiv.org/abs/2607.09816

机构：Columbia University（哥伦比亚大学）； University of California, Los Angeles（加利福尼亚大学洛杉矶分校）； University of Melbourne（墨尔本大学）

作者：Yanxuan Yu, Dong liu, Renata Borovica-Gajic, Ying Nian Wu

英文摘要：Class imbalance poses a fundamental challenge in risk-sensitive applications such as fraud detection and medical diagnosis, where minority-class samples are scarce yet critical for accurate classification. Existing oversampling methods generate synthetic samples to rebalance class distributions; however, they often produce large numbers of low-quality candidates that distort decision boundaries or introduce artifacts, leading to overfitting and degraded generalization. In this work, we introduce RUBRIC, a generator-agnostic filtering framework that formulates synthetic sample selection as a quality-over-quantity optimization problem. RUBRIC ranks candidates using a realism-utility trade-off: realism is quantified by a learned discriminator that distinguishes real samples from synthetic samples, while utility captures proximity to the decision boundary through a concave margin-based scoring function. We show that, under mild regularity conditions, the proposed filtering strategy monotonically tightens the generalization bound for margin-based classifiers by jointly reducing distribution shift and suppressing near-negative tail contributions. Through extensive experiments on credit-card fraud detection and other imbalanced benchmarks, we demonstrate that RUBRIC improves F1-macro and recall while maintaining comparable ROC-AUC across several generators. We also provide explicit lambda-sensitivity analysis to show how users can recover AUPRC when ranking quality is prioritized.

AI 总结：该研究针对具有面板数据的马尔可夫链选择模型，提出包含参数估计、选择预测及商品组合优化的框架。通过纳入偏序偏好信息的新颖期望最大化算法进行参数估计，在数据集上表现优于传统算法，还给出相关问题的难度与计算结果。

链接：https://arxiv.org/abs/2607.09817

机构：Melbourne Business School（墨尔本商学院）

作者：Yalcin Akcay, Gerardo Berbeglia, Young-San Lin

英文摘要： We propose a framework for the Markov chain (MC) choice model with panel data, including parameter estimation, personalized choice prediction, and personalized assortment optimization. In contrast to the traditional setting, which assumes that each transaction is independently drawn from a random utility model, our framework accounts for dependencies among transactions for the same customer in historical data, captured by partial-ordering preference information. To the best of our knowledge, our framework initiates the study of choice modeling with panel data under MC. As our primary result, we propose novel expectation-maximization (EM) algorithms for MC parameter estimation by incorporating partial-ordering-based customer preference information. On synthetic datasets and the sushi dataset, our EM algorithms outperform the traditional EM algorithm of Simsek and Topaloglu (Operations Research, 66, 2018) and multinomial-logit-based partial-order benchmarks adapted from Jagabathula and Vulcano (Management Science, 64, 2018). As our secondary contribution, we present hardness and computational results for conditional choice prediction and assortment optimization problems. These results complement our estimation framework and clarify the computational landscape of conditional choice and assortment optimization, which may be of independent interest.

AI 总结：研究针对预测后优化系统的不足，提出学习预测模糊集方法，通过深度上下文模型输出相关参数定义模糊集，经特定训练使半径自适应，用于分布鲁棒投资组合优化，效果优于多种基线，能减少保守性并提高适应性。

链接：https://arxiv.org/abs/2607.09820

作者：Junjie Guo

英文摘要：Predict-then-optimize systems usually compress uncertainty into a point forecast and then solve a downstream optimization problem as if the forecast were reliable. Distributionally robust optimization (DRO) offers protection against misspecification, but the ambiguity set is often centered at historical samples and uses a fixed radius. We propose \emph{learned predictive ambiguity sets} (LPAS): a deep contextual model outputs a finite nominal scenario distribution, a state-dependent Wasserstein radius, and optionally an anisotropic ground metric. These outputs define a contextual ambiguity set that feeds a DRO decision layer. The radius is trained by a combination of conditional quantile calibration, size regularization, and downstream decision loss, so that robustness is adaptive rather than globally fixed. We derive the finite dual form used by the decision layer, present a staged training algorithm, and evaluate the method on distributionally robust portfolio optimization with 20 S&P 500 constituents from 2018--2026. The proposed method substantially improves over equal-weight, predict-then-optimize, and historical Wasserstein DRO baselines, achieving 26.28% annualized return, Sharpe ratio 1.30, final wealth 1.61, and lower tail loss than a deep fixed-radius DRO baseline while using a smaller average radius. The results show that learned ambiguity radii can recover most of the performance of strong fixed-radius DRO while reducing unnecessary conservatism and improving regime adaptivity.

AI 总结：研究加拿大动态食品定价中消费者被剥削风险，提出将公平约束嵌入零售销售预测的方法，用ARDL模型结合线性规划和模拟退火解决定价问题，对比多种基准模型验证效果，构建出注重公平的定价框架。

链接：https://arxiv.org/abs/2607.09956

机构：The University of Winnipeg（温尼伯大学）

作者：Sujay Uday Rittikar

英文摘要：Pricing food products to balance profitability with consumer welfare is a central challenge for retailers. Dynamic pricing is widely used to maximize revenue, yet most pricing models optimize business objectives while overlooking consumer fairness. This paper studies the risk of consumer exploitation under dynamic food pricing in Canada and proposes a methodology that embeds fairness constraints directly into retail sales forecasting. We model total retail trade sales with a log--log Autoregressive Distributed Lag (ARDL) specification, in which the coefficient on a product price is a sales elasticity, and pose the pricing problem as maximizing forecast sales subject to price bounds anchored to the Consumer Price Index (CPI). We solve this problem with both Linear Programming (LP) and Simulated Annealing (SA), under single-product and multi-product configurations. A key finding is that the fitted nominal elasticities are positive. As a result, an unconstrained sales-maximizer would push every price to its upper bound, and the CPI ceiling is the safeguard that prevents this. Simulated Annealing instead settles on conservative, interior prices that lower consumer cost while still meeting the sales target. We benchmark forecast accuracy against naive, seasonal-naive, ARIMA, and SARIMA baselines, and a CPI-deflated re-specification shows that the positive nominal elasticities are largely an inflation-driven artifact. The result is a transparent, fairness-aware pricing framework.

AI 总结：研究针对神经网络操作乘法性质及现有优化器问题，提出指数线性权重重参数化方法\method，结合对称指数与线性路径，经实验验证其能提升损失下降速度，在变压器训练中减少训练步骤，还有不匹配初始化可改善早期优化。

链接：https://arxiv.org/abs/2607.09967

机构：Canva Research（Canva研究公司）

作者：Ethan Smith

英文摘要：Many neural networks operations have a multiplicative nature rather than additive: halving or doubling a norm are analogous relatively but require unequal optimization distances when taking linear steps. Adaptive optimizers such as Adam normalize updates per coordinate, but update steps remain additive; weights with very different magnitudes receive similarly sized absolute changes, producing very different relative perturbations. We introduce \textbf{\method} (\textbf{\methodshort}), a weight reparameterization for neural networks that combines a sign-aware symmetric-exponential pathway with an identity-like linear pathway. The symmetric-exponential pathway is near-linear for small raw weights but increasingly curved at larger magnitudes. Additive updates in logarithmic space map to magnitude-proportional changes in effective weight space. The linear pathway provides a direct route through the transform that we hypothesize stabilizes optimization, while learnable scale, curvature, and offset parameters control balance between pathways and the curvature of the exponential pathway. These components create a curved parameter-space geometry that empirically improves speed of loss descent over standard linear parameterization. We also identify a useful \emph{mismatched initialization}: raw weights are chosen so a symmetric version of the transform matches Xavier statistics, but training uses an asymmetric forward transform that leaves positive weights at full strength while making negative weights smaller in magnitude; in small-model ablations, this improves early optimization and may act as a form of symmetry breaking. We train transformers on OpenWebText over nine width$\times$depth configurations, \methodshort reaches matched validation loss in 1.32--1.49$\times$ fewer training steps, with the largest widths seeing the biggest gains.

AI 总结：研究虚假相关性下数据不平衡对模型泛化的影响，通过改变虚假比率和模型容量发现数据不平衡能促进强大模型泛化，如合成任务中2层变压器在特定虚假比率下泛化效果提升，还通过机制分析找到相关途径。

链接：https://arxiv.org/abs/2607.10116

机构：University of California, Los Angeles（加利福尼亚大学洛杉矶分校）； Purdue University（普渡大学）

作者：Cheng-Ting Chou, Duc Binh Hoang

英文摘要：We study robust generalization under spurious correlations: tasks where a shortcut feature is correlated with the true label in training but anti-correlated in an adversarial held-out split. Varying the spurious ratio $r$ (the fraction of training examples where shortcut = true label) and model capacity, we find a counterintuitive result: data imbalance promotes generalization in sufficiently capable models. On a synthetic task where the true label is sum parity of an integer sequence and the shortcut is the parity of the maximum-valued element, a 2-layer, 2-head transformer generalized (reached $100\%$ adversarial accuracy) in 0% of seeds at $r{=}0.50$ but 77% of seeds at $r{=}0.90$. The effect is absent in 1-layer models, where imbalance instead traps the model on the shortcut. Through mechanistic analysis -- gradient conflict dynamics, circuit evolution, and QK/OV circuit ablations -- we characterize a mechanistic pathway consistent with imbalance promoting generalization.

AI 总结：研究电信公司客户流失问题，提出集成机器学习流失预测、客户细分及定制营销和ROI策略的数据驱动框架，用特定数据集训练调整模型，划分客户聚类并设计策略，经实验验证该框架比单独流失预测能产生更优营销决策。

链接：https://arxiv.org/abs/2607.10260

作者：Nada Ali, Lina Ahmed, Tahani Abdalla Attia Gasmalla

英文摘要：Customer churn is a major challenge for telecommunication companies, directly eroding revenue and long term customer relationships. Traditional retention programs rely on generic, not personalized incentives and lack the precision to identify high risk customers before they leave. This paper presents a data driven marketing optimization framework integrating machine learning based churn prediction, customer segmentation combining churn risk with customer value, and tailored, segment specific marketing and Return on Investment ROI strategies. Using the IBM Telco Customer Churn dataset with 7043 customers and 21 features, three gradient boosting ensembles, XGBoost, LightGBM, and CatBoost, were trained and tuned via randomized search with stratified 5 fold cross validation, class weighting, and F1 score driven decision threshold optimization to counter a class imbalance of 73.4% versus 26.6%. CatBoost was selected as the deployment model, achieving 77.68% accuracy, an F1 score of 0.6366, a PR AUC of 0.6553, and a ROC AUC of 0.8403 on the held out test set. Customers were partitioned with K Means clustering, validated via the Elbow method and visualized with Principal Component Analysis, into High, Medium, and Low Value segments, cross tabulated against churn risk labels to define four actionable clusters. Segment specific retention, upsell, and engagement strategies were designed for each cluster, and a theoretical ROI and CLV framework quantifies the financial impact of the proposed interventions. The pipeline was operationalized in an interactive Streamlit web application allowing marketing teams to upload data, filter by segment, visualize churn drivers via SHAP, and download automated segment reports. Results confirm that combining predictive churn modeling with value aware segmentation yields more actionable and profitable marketing decisions than churn prediction alone.

AI 总结：研究双层优化中理论与实际单环实现的差距，利用解耦范数分析框架，改进单环近似隐式微分和迭代微分方法的收敛结果，提升收敛速率并精确匹配渐近误差下界，经实验验证理论发现。

链接：https://arxiv.org/abs/2607.10263

机构：School of Mathematics and Statistics, Xi’an Jiaotong University（西安交通大学数学与统计学院）； School of Data Science, Fudan University（复旦大学数据科学学院）； SGIT AI Lab, State Grid Corporation of China（国家电网公司SGIT人工智能实验室）； Center for Intelligent Decision-Making and Machine Learning, School of Management, Xi’an Jiaotong University（西安交通大学管理学院智能决策与机器学习中心）

作者：Yubo Zhou, Jun Shu, Luo Luo, Junmin Liu, Deyu Meng, Guang Dai, Haishan Ye

英文摘要：Bilevel optimization underpins many machine learning applications, including hyperparameter optimization, meta-learning, neural architecture search, and reinforcement learning. While hypergradient-based methods have advanced significantly, a gap persists between theoretical guarantees and practical single-loop implementations required for efficiency. We bridge this gap by establishing sharper convergence results for single-loop approximate implicit differentiation (AID) and iterative differentiation (ITD) methods, leveraging our proposed analytical framework, decoupled norm analysis (DNA). For AID, we improve the convergence rate from $\mathcal{O}(κ^6/K)$ to $\mathcal{O}(κ^5/K)$, where $κ$ is the condition number of the inner-level problem. For ITD, we prove that the asymptotic error is $\mathcal{O}(κ^2)$, exactly matching the known lower bound and improving upon the previous $\mathcal{O}(κ^3)$ guarantee. Numerical experiments on synthetic and real tasks corroborate our theoretical findings.

AI 总结：研究流形上束值统计量，通过希尔伯特空间不等式推导集中界，分离偏差与方差，给出精确公式，建立稳健估计器，经实验验证理论预测。

链接：https://arxiv.org/abs/2607.10592

机构：Indian Statistical Institute（印度统计研究所）； VSB Technical University of Ostrava（俄斯特拉发技术大学）

作者：Swagatam Das, Vaclav Snasel

英文摘要： Many geometric statistics and manifold learning pipelines routinely produce observations -- such as tangent vectors or local frames -- whose natural home is a varying family of fibers attached to different points of a base manifold, rather than a single shared vector space. Forming empirical averages requires transporting these observations to a common reference fiber, thereby introducing curvature- and holonomy-driven effects that are absent from classical concentration theory. We develop a non-asymptotic concentration theory for such transported empirical means, deriving finite-sample, dimension-free Hoeffding- and Bernstein-type bounds via sharp Hilbert-space inequalities. When shortest paths to the reference point are non-unique, transport becomes path-dependent and introduces a deterministic holonomy bias; we isolate and quantify this bias through bundle curvature and loop geometry, with sharp closed-form formulas for the tangent bundle of a round sphere. The resulting bias-variance decomposition separates the stochastic fluctuation decaying at the classical $n^{-1/2}$ rate in sample size $n$, from a curvature-driven error floor that no amount of additional data can eliminate; minimax lower bounds confirm both terms are unavoidable. We further establish a robust median-of-means estimator achieving optimal rates under heavy tails and the central limit theorem in the reference fiber. Controlled experiments on the sphere validate all theoretical predictions.

AI 总结：研究低精度训练中标准优化器的问题，提出结合加法和乘法更新的M+Adam方法，经证明在标准假设下单调下降，在多种模型预训练中持续改善低精度训练。

链接：https://arxiv.org/abs/2607.10611

机构：California Institute of Technology（加利福尼亚理工学院）； University of Copenhagen（哥本哈根大学）； Aarhus University（阿鲁斯大学）

作者：Xiaoyuan Liang, Sebastian Loeschcke, Mads Toftrup, Anima Anandkumar

英文摘要：Training with quantized weights can reduce costs but often results in degraded accuracy, especially when optimization is carried out in low precision, without storing high-precision copies. We identify a key failure mode: under low precision, standard optimizers can get stuck and not make progress, especially at large weight magnitudes due to coarse mantissa resolution. To overcome this, multiplicative updates have been previously proposed, in place of additive updates in standard optimizers. While successful under extremely low precision, such as under the logarithmic number system, they suffer from failures near zero and across sign changes. The failure modes of additive and multiplicative updates are therefore complementary. To exploit this, we propose M+Adam, which combines both update types: additive steps handle sign changes and small magnitudes, while multiplicative steps ensure progress at large magnitudes when additive updates are zeroed out under rounding. We prove monotone descent for M+Adam under standard smoothness assumptions. Across LLaMA-style pretraining with 60M-1B models, 1x-8x Chinchilla budgets, and using only BF16, FP8, and FP4 master weights, M+Adam consistently improves low-precision training.

AI 总结：针对实际异方差和非平稳问题，引入受HEBO启发的tidyHEBO模型，重构其设计理念并修订多方面内容。经多任务基准测试，该模型性能具竞争力且鲁棒性强，可作顺序实验实用工具和贝叶斯优化研究通用基准。

链接：https://arxiv.org/abs/2607.10669

作者：L. A. Zhukov, E. V. Shaburova, D. V. Antonets

英文摘要：Bayesian optimization is increasingly used to guide data-efficient experimentation in chemistry, materials science, and related laboratory settings, but its practical performance depends strongly on how well surrogate-model assumptions match the geometry and noise structure of the underlying objective. We introduce tidyHEBO, a robust Bayesian optimization model inspired by heteroskedastic evolutionary Bayesian optimization (HEBO) for single-objective, sequential optimization. tidyHEBO reconstructs the HEBO design philosophy in BoTorch and revises surrogate training, output-warping selection, acquisition function evaluation, and Pareto-front search. We benchmarked tidyHEBO on synthetic functions, Olympus emulators, fully experimental reaction-optimization datasets, needle-in-a-haystack (NIAH) materials problems, and Bayesmark hyperparameter optimization tasks. On these tasks tidyHEBO achieved competitive to superior performance and improvement in robustness across repeated optimization runs. We therefore propose tidyHEBO as a practical tool for sequential experimentations and a strong general-purpose benchmark for future Bayesian optimization research.

AI 总结：研究预层归一化循环变压器中层归一化作用，指出其为隐式增益控制，分析稳定性预算及进位约束，通过六个任务实验表明线性进位非深度记忆机制，刻画该说法边界，结果经解析得出并在CPU规模实现中验证。

链接：https://arxiv.org/abs/2607.10681

机构：Faculty of Business and Economics, The University of Hong Kong（香港大学经管学院）

作者：Matthias M. M. Buehlmaier

英文摘要：In pre-LayerNorm looped transformers, LayerNorm inside the recurrent block acts as an implicit gain controller: by coupling the block's local Lipschitz constant inversely to the activation scale, it renders the recurrence Jacobian non-normal -- asymptotically contractive at every verified fixed point even where its operator norm exceeds 1 -- so the true stability budget is the spectral margin, not an operator-norm bound. That margin depletes as the carry $ρ\to 1$, and a minority of initializations never converge to a fixed point at all, so the diagonal carry constraint $ρ(\bar{A}) < 1$ is necessary but not sufficient for convergence of the full recurrence. Training experiments across six tasks, including a controlled ablation, reveal that the linear carry is not the depth-memory mechanism: gradient descent routes memory through the block's more expressive nonlinear recurrence and leaves the stability-constrained carry at rest -- the carry's role is stabilization, not memory. We characterize the boundary of this claim: on tasks with axis-aligned per-channel structure, gradient descent does recruit the carry. All results are derived analytically and verified in a from-scratch, CPU-scale implementation; verification at larger scale is needed.

AI 总结：研究实巴拿赫空间中扩展球等几何PCC的VC维数，通过映射线性化距离，利用平衡符号和估计等方法，得到\(L_p(\mu)\)中扩展球PCC的VC维数无维数上界及匹配下界，还推导了密集邻域引理。

链接：https://arxiv.org/abs/2607.10751

机构：Pontifícia Universidade Católica do Rio de Janeiro（里约热内卢天主教大学）； Alfréd Rényi Institute of Mathematics（阿尔弗雷德·雷尼数学研究所）； Loránd Eötvös University（罗兰·厄特沃什大学）

作者：Grigory Ivanov, Attila Jung, Márton Naszódi

英文摘要： Following Alon, Hanneke, Holzman, and Moran (FOCS 2021), we define a partial concept class (PCC) as a family of partial functions \(f: V\to\{0,1,\ast\}\); equivalently, its concepts partition the ground set into black ($f^{-1}(1)$), grey ($f^{-1}(\ast)$), and white parts ($f^{-1}(0)$). Its VC dimension is defined by shattering sets on which the value $\ast$ is not taken. We study two geometric PCCs in real Banach spaces, both with a margin \(δ>0\): expanded half-spaces, where the grey part is a strip of width at least \(δ\) adjacent to a half-space, and expanded balls, where the grey part is an annulus of width \(δ\) around a unit radius ball. Our main results are dimension-free upper bounds on the VC dimension of the PCC of expanded balls in \(L_p\parenthμ\), \(1\le p

AI 总结：研究约束在线凸优化问题，针对当前最好的OGD+投影算法，给出其累积约束违反的下界为\(\Omega (T^{\frac{d - 1}{2d}})\)，此为该问题的首个下界结果。

链接：https://arxiv.org/abs/2607.10808

作者：Haricharan Balasundaram, Karthick Krishna Mahendran, Rahul Vaze

英文摘要：The problem of constrained online convex optimization is considered, where at each round, once a learner commits to an action $x_t \in \mathcal{X} \subset \mathbb{R}^d$, a convex loss function $f_t$ and a convex constraint function $g_t$ that drives the constraint $g_t(x)\le 0$ are revealed. The objective is to simultaneously minimize the static regret and cumulative constraint violation (CCV) compared to the benchmark that knows the loss functions and constraint functions $f_t$ and $g_t$ for all $t$ ahead of time, and chooses a static optimal action that is feasible with respect to all $g_t(x)\le 0$. Currently, the best known algorithm is OGD+Projection algorithm of [Vaze and Sinha, 2025] that has simultaneous regret of $O(\sqrt{T})$ and CCV of $O(T^{1/3})$ for $d=2$ [Balasundaram et al., 2026], and simultaneous regret of $O(\sqrt{T})$ and CCV of $O(\sqrt{T})$ for any $d$ [Sarkar and Sinha, 2026]. In this paper, we show that the CCV of the OGD+Projection algorithm is $Ω(T^{\frac{d-1}{2d}})$. This is the first such lower bound result.

AI 总结：研究大语言模型中Transformer动力学集体行为的物理机制，利用Pythia语言模型提取弛豫谱，定量测量相关指标，发现其动力学受红外集体组织支配，红外慢模式组织是普遍集体原则。

链接：https://arxiv.org/abs/2607.10923

机构：Electronics and Telecommunications Research Institute（电子和电信研究所）

作者：Byung Gyu Chae

英文摘要：Large language models exhibit remarkable emergent behaviors, yet the physical mechanism governing their collective dynamics remains poorly understood. Cognitive Field Theory predicts that learning reorganizes the time-scale density of states (TDOS) through the infrared accumulation of slow relaxation modes, thereby enhancing the memory self-energy, reducing the cognitive forgetting gap, and strengthening the collective susceptibility. Using publicly available Pythia language models, we extract relaxation spectra directly from Transformer layer Jacobians throughout training, network depth, and model scale, allowing the TDOS, memory self-energy, forgetting gap, memory kernel, and infrared critical exponent to be measured quantitatively. The measurements reveal progressive infrared accumulation of slow relaxation modes, producing an approximately flat infrared TDOS with \( ρ(λ)\simλ^{-0.1} \) and scale-free memory kernels \( K(t)\sim t^{-1}. \) The memory self-energy exhibits a pronounced transient maximum during early optimization before relaxing toward a metastable near-critical regime, corresponding to the smallest cognitive forgetting gap and the largest collective susceptibility predicted by Cognitive Field Theory. These observations provide quantitative experimental evidence that Transformer dynamics are governed by infrared collective organization. The reproducibility of the same dynamical behavior across training, network depth, and model scales suggests that infrared slow-mode organization represents a universal collective principle of Transformer dynamics.

AI 总结：研究强盗反馈版本的在线主成分分析，改进了遗憾值的上下界，弥合差距至\(r\sqrt{dT}\)。上界通过新算法实现，结合在线镜像下降与多尺度探索；下界通过构造自适应对手，将遗憾值下界估计转化为子空间估计问题，并讨论其与量子断层扫描联系。

链接：https://arxiv.org/abs/2607.10936

机构：Georgia Tech（佐治亚理工学院）； UIC（伊利诺伊大学芝加哥分校）

作者：Moïse Blanchard, Dmitrii Ostrovskii, Aadirupa Saha

英文摘要：We study the bandit-feedback version of online principal component analysis (Bandit PCA): in each round $t = 1,\dots,T$, the adversary selects a $d \times d$ symmetric gain matrix $G_t$ with spectrum in $[0,1]$ and rank at most $r$; the learner simultaneously selects a unit vector $w_t \in S^{d-1}$ and receives the reward $w_t^\top G_t w_t$. The learner receives no other feedback, and aims to minimize the regret against the best unit vector in hindsight. This problem was introduced by Kotlowski and Neu (2019), who gave an algorithm with regret $O(d\sqrt{rT \log T})$ and showed the lower bound of $Ω(r\sqrt{T/\log T})$. We improve upon both of these bounds and essentially bridge the gap between them, establishing the minimax regret of order $r\sqrt{dT}$ up to polylogarithmic factors in $d$ and $T$. The upper bound is attained by a novel algorithm, which combines online mirror descent on the spectrahedron of (real) density matrices with a multiscale exploration scheme in which the eigenspaces with different spectral magnitudes are updated at different rates. For the lower bound, we construct an adaptive adversary that refines a hidden large-reward subspace based on the learner's actions, in such a way that low regret is impossible without estimating the subspace; as a result, lower-bounding the regret reduces to studying the arising subspace estimation problem. Finally, we discuss connections of Bandit PCA with adaptive-measurement quantum tomography.

AI 总结：研究提出WSqD学习率调度方法，受随机凸优化启发，用移位反平方根基础取代WSD恒定稳定阶段，保留最终线性冷却，其基础学习率调度与训练轮次无关且收敛速率最优，实验显示在语言模型预训练中表现良好。

链接：https://arxiv.org/abs/2607.10959

机构：University of Pennsylvania（宾夕法尼亚大学）

作者：Jianhao Ma, Yuxin Chen

英文摘要： Standard learning rate schedules such as cosine annealing are tied to a fixed training horizon, limiting their ability to accommodate post hoc horizon extension. Warmup-stable-decay (WSD) partially addresses this issue by maintaining a long constant-rate phase before a short linear cooldown, allowing training to resume from a pre-decay checkpoint. However, its peak learning rate is still tuned based on the original training horizon and can become suboptimal when training is extended. Motivated by stochastic convex optimization, we propose WSqD (Warmup with Square-root base and linear Decay), a learning rate schedule that replaces WSD's constant stable phase with a shifted inverse-square-root base while retaining the final linear cooldown. In the stochastic convex setting, WSqD provably attains the minimax-optimal $O(1/\sqrt{T})$ last-iterate convergence rate. Importantly, its base learning rate schedule is horizon-independent, and the training horizon is needed only to determine when to begin the final cooldown. Empirically, on language-model pretraining using the SlimPajama corpus, WSqD matches or outperforms carefully tuned WSD and other baselines across multiple training horizons while reusing a single peak learning rate.

AI 总结：本文针对复杂零维降阶模型规划中依赖人工、传统方法局限等问题，提出多智能体架构Z-COPA，其核心是专用图表示法，将规划转化为图结构优化问题，经多基准测试验证，该框架性能卓越且打破传统范式，提供新规划技术方法。

链接：https://arxiv.org/abs/2607.10994

机构：Institute of Engineering Thermophysics, Chinese Academy of Sciences（中国科学院工程热物理研究所）； National Key Laboratory of Science and Technology on Advanced Light-duty Gas-turbine（先进轻型燃气轮机技术重点实验室）； University of Chinese Academy of Sciences（中国科学院大学）； Qingdao Institute of Aeronautical Technology（青岛航空技术研究院）； Nanjing Future Energy System Research Institute（南京未来能源系统研究院）

作者：Bingteng Sun, Hao Yin, Yiling Chen, Renjie Xiao, Lei Xie, Shanyou Wang, Ruonan Wang, Shubao Chen, Qingzong Xu, Lin Lu, Qiang Du, Junqiang Zhu

英文摘要：Zero-dimensional reduced-order models (0D ROMs) are central to multi-dimensional design workflows for high-end complex equipment. However, the planning process currently relies on manual expertise, limiting topological exploration and prolonging iterations. Even traditional optimization methods such as Genetic Algorithms (GA) are typically confined to local parameter tuning. Although Large Language Model (LLM) agents have shown promise in exploring large sample spaces, and frameworks such as Chain of Thought (CoT) and Reason and Act (ReAct) improve reasoning reliability, while Retrieval-Augmented Generation (RAG) overcomes domain knowledge barriers, a single agent still falls short for the long-horizon and highly coupled nature of complex 0D ROM planning. This paper proposes the Zero-dimensional reduced-order model CO-Planning framework (Z-COPA), a multi-agent architecture featuring a Symbolic Action Graph Engine (SAGE) and a MILP-Guided Navigation (MGN) optimizer. Its core innovation is a dedicated graph representation method that accurately encodes the 0D flow network topology, converting the empirical planning process into a rigorous graph structure optimization problem. We validate the forward and inverse design capabilities and generalization performance of Z-COPA on two real aircraft engine secondary-air systems, two IEEE power-distribution reconfiguration benchmarks, and two water-distribution network benchmarks. The results show superior task completion quality, obtaining the best performance in both forward and reverse design of air systems. Z-COPA disrupts the traditional 0D model planning paradigm, providing a new technical approach for exploring broader topological space and achieving highly automated, globally optimal air system architectures.

AI 总结：研究针对物理信息神经网络同步优化与物理传播路径不符问题，基于物理信息传播路径定义训练优先级，用NTK动力学分析原因，提出统一多维优先级约束框架，改善了PINNs在特定问题上的收敛行为和预测准确性。

链接：https://arxiv.org/abs/2607.11094

作者：Zhangyi Lian, Xinda Dong, Wenxuan Huo, Weifeng Huang, Gang Zhu, Qiang He

英文摘要：Physics-informed neural networks (PINNs) have shown promise for solving partial differential equations (PDEs); however, their synchronous optimization treats residuals of different regions and constraints equally, which is inconsistent with the progressive "from source to response" physical information propagation path, degrading training stability and accuracy. Existing causal training methods focus mainly on the temporal dimension, lacking a unified characterization of spatial and boundary dimensions. To address this, we define a unified class of training priorities according to the physical information propagation path: premise regions should be learned before dependent regions; temporal, spatial, and boundary priorities are instances of this principle. Using neural tangent kernel (NTK) dynamics, we theoretically analyze why standard PINNs do not obey this priority: their residual convergence order is governed by the NTK spectrum and is independent of the propagation path. Accordingly, we propose a unified multi-dimensional priority-constraint framework that partitions the domain along the propagation path and constructs negative-exponential residual weights, converting the physical propagation order into a training priority. For cases with coexisting priorities, we introduce a directional compatibility coefficient to clarify that "orthogonal directions can be coupled multiplicatively in synergy, whereas coaxial opposite directions cannot." Benchmark cases show that this method consistently improves the convergence behavior and prediction accuracy of PINNs on problems with clear propagation paths or constraint-dominated structures, without modifying the network architecture and with controllable additional computational cost.

AI 总结：研究针对大语言模型代理依赖外部工具时工具知识难共享问题，提出ToolAtlas框架，通过执行验证探测构建提供商端工具内存，经实验证明该框架能有效提升性能且可跨环境和框架复用，确立了提供商端工具内存范式。

链接：https://arxiv.org/abs/2607.11126

机构：School of Computer Science, Peking University（北京大学计算机科学学院）； Microsoft（微软）

作者：Yue Fang, Zhibang Yang, Fangkai Yang, Xiaoting Qin, Liqun Li, Qingwei Lin, Saravan Rajmohan, Dongmei Zhang

英文摘要： Large language model (LLM) agents increasingly rely on external tools served by shared providers and accessed by heterogeneous downstream agents. Existing approaches improve tool use on the agent side through parameter updates, prompt refinement, or agent-side memory, making tool knowledge difficult to share and limited to behaviors observed in past tasks. We argue that reusable tool knowledge should instead be maintained by the tool provider. We introduce ToolAtlas, a graph-based framework that builds a persistent provider-side tool memory of tool capabilities, failure boundaries, and cross-tool compositions through execution-verified probing. At inference time, agents query the tool memory via adaptive graph traversal. Across two MCP-based benchmarks spanning eight services, ToolAtlas outperforms existing tool-side optimization and agent-side memory baselines by up to 21.61% in pass@1 and 18.61% in pass@4. The same tool memory also transfers across environment instances and agent frameworks without retraining or task-time exploration, yielding up to 24.16%/16.22% and 17.49%/14.27% relative gains in pass@1/pass@4, respectively. Ablation studies show that these gains arise from combining tool-centered memory organization with capability-guided execution probing. These results establish provider-side tool memory as an effective and reusable paradigm for tool servers. Our code is in: https://github.com/PuppyKnightUniversity/ToolAtlas.

AI 总结：本文提出NeuroMem-FHP框架，用LSTM网络和Transformer直接从到达间隔时间序列估计分数阶霍克斯过程参数，无需似然优化。实验表明其优于经典MLE方法，在真实数据集上也表现良好，为FHP参数估计提供了准确高效的替代方法。

链接：https://arxiv.org/abs/2607.11177

机构：Indian Institute of Management Indore（印度管理学院印多尔分校）

作者：Neha Gupta, Aditya Maheshwari

英文摘要：In this paper, we propose deep learning based NeuroMem-FHP framework for estimating the parameters of the fractional Hawkes process (FHP), a self-exciting point process that captures long-range dependence through a fractional Mittag-Leffler excitation kernel. Two neural architectures, namely a Long Short-Term Memory (LSTM) network and a Transformer, are developed to estimate the model parameters $(μ,γ,α,β)$ directly from sequences of inter-arrival times without requiring computationally intensive likelihood optimization. Experiments on synthetic data that both neural models significantly outperform the classical Maximum Likelihood Estimation (MLE) method, with the Transformer achieving the highest estimation accuracy (MSE = $0.1634$), followed by the LSTM (MSE = $0.1752$), compared to MLE (MSE = $2.8032$). An ablation study further examines the effects of key hyperparameters on model performance. The proposed framework is also on two real-world high-frequency datasets, namely AAPL NBBO transaction data and Montgomery County 911 emergency call records. Using a predictive validation approach, event sequences simulated from the estimated parameters closely reproduce the empirical distribution, tail behavior, and temporal dependence structure of the observed data. These results demonstrate that Transformer-based parameter estimation provides an accurate and efficient alternative to conventional estimation techniques for FHP and offers a promising framework for modeling event-driven systems with long-memory dynamics.

AI 总结：研究智能体技能优化问题，提出LASKO框架，将技能建模为李代数胚截面，利用李括号筛选测试替代昂贵验证，在自然语言任务因果提取中实现近15倍加速，提升技能优化速度。

链接：https://arxiv.org/abs/2607.11493

机构：Adobe Research（Adobe研究院）； University of Massachusetts, Amherst（马萨诸塞大学阿默斯特分校）

作者：Sridhar Mahadevan

英文摘要：Agentic systems increasingly improve themselves by editing skills: prompts, rubrics, plans, tool contracts, examples, validators, and traces. Skill edits are not independent coordinates in a vector space: they are local repairs to structured artifacts whose effects are observed only after rollout, validation, and critique. Distinct edits can have the same immediate visible effect while differing in routing context, template state, guardrail scope, or future composability. The order of edits can matter as well: repairing a schema before a normalization rule need not be equivalent to applying the same edits in the reverse order. This paper introduces a new framework for skill optimization called LASKO, for Lie Algebroid SKill Optimization. LASKO models typed, anchored Markdown skills as the base category and available edit policies as sections of a controlled Lie algebroid with anchor $ρ$. The anchor maps an edit policy to its visible Markdown effect; the kernel $\ker(ρ)$ represents latent template, routing, or implementation structure; and the algebroid bracket measures noncommuting edit composition. As shown in the paper, LASKO achieves order-of-magnitude speedups in skill optimization in our preliminary benchmark results, primarily because it substitutes inexpensive Lie-bracket screening tests that run in microseconds, before investing in expensive validations that require running large language models. On a causal extraction from natural language task, LASKO achieved a speedup of almost $15 \times$ compared to a brute-force approach that validated all edits by running them through a DeepSeek V3.1 4-bit model with 671B parameters.

AI 总结：研究针对大语言模型训练后优化信号耦合问题，提出PUST框架，通过代理模型探索、提取并传输相对改进信号，解耦更新信号探索与分布对齐，减少计算开销，支持异步处理与跨模型转移，提升训练后范式的模块化、可重用性与效率。

链接：https://arxiv.org/abs/2607.11505

作者：Daocheng Fu, Rong Wu, Yu Yang, Xuemeng Yang, Jianbiao Mei, Licheng Wen, Pinlong Cai, Yong Liu, Botian Shi, Yu Qiao

AI 总结：研究针对因果发现中特定数据集方法的局限，提出通用框架CDFM，通过研究因果可识别性理论边界构建变分框架，将未知因果机制视为潜在变量，经大量合成模型预训练，性能优于传统算法，推动因果发现范式转变。

链接：https://arxiv.org/abs/2607.11508

机构：School of Computer Science, Guangdong University of Technology, Guangzhou, China（广东技术大学计算机科学学院）； College of Mathematics and Computer, Shantou University, Shantou, China（汕头大学数学与计算机学院）； Department of Computer Science and Technology, Tsinghua University, Beijing, China（清华大学计算机科学与技术系）

作者：Jie Qiao, Ruichu Cai, Zijian Li, Weilin Chen, Pengfei Hua, Boyan Xu, Zhengming Chen, Zhifeng Hao, Peng Cui

AI 总结： 研究针对MatBench缺乏反映AI预测模型在分布外材料数据上性能能力的问题，结合MatBench流程与OOD性能评估研究，提出自动工具包AutoMatBench，经实验验证其能节省成本并获相似结果，对新材料发现有积极贡献。

链接：https://arxiv.org/abs/2607.11526

机构：Institute of Computing Technology, Chinese Academy of Sciences, Beijing, China（中国科学院计算技术研究所）； University of Chinese Academy of Sciences, Beijing, China（中国科学院大学）

作者：Hongxiao Li, Wanling Gao

AI 总结：该研究提出用随机标签预测头研究深度神经网络分类任务中的记忆，通过其性能估计拉德马赫复杂度，分析泛化与过拟合，还基于此提出正则化技术，挑战了过拟合等同于记忆的传统假设，揭示减少记忆对泛化的复杂影响。

链接：https://arxiv.org/abs/2607.11541

机构：University of Münster（明斯特大学）

作者：Marlon Becker, Jonas Konrad, Luis Garcia Rodriguez, Benjamin Risse

AI 总结：研究神经集函数学习，针对现有最优子集预言机框架依赖蒙特卡罗采样估计梯度导致计算开销大且轨迹不稳定的问题，提出将证据下界重新解释为连续松弛并学习替代目标，实验证明该方法能减少开销、加速推理并优于现有基线。

链接：https://arxiv.org/abs/2607.11555

机构：Fuzhou University（福州大学）； Imperial College London（伦敦帝国学院）； National University of Singapore（新加坡国立大学）

作者：Yongquan Shi, Zijing Ou, Shiping Wang, Yatao Bian

AI 总结：提出协作元学习框架用于分布式贝叶斯优化，无需原始数据交换。揭示梯度共享问题，评估差分隐私防御，刻画其隐私 - 效用权衡。

链接：https://arxiv.org/abs/2607.11600

机构：School of Industrial Engineering and Management（工业工程与管理学院）； Oklahoma State University, Stillwater, OK, USA（俄克拉荷马州立大学）

作者：Aditya Rane, Sathwik Yamana, Paritosh Ramanan, Srikanthan Ramesh, Akash Deep

AI 总结：研究神经网络‘顿悟’现象中表示几何与延迟泛化的关系，引入几何维度正则化方法，发现其能改变‘顿悟’动态，加速泛化，为研究和影响神经网络延迟泛化提供实用途径。

链接：https://arxiv.org/abs/2607.11666

作者：Maksim A Kazanskii

AI 总结：研究针对硬件感知神经网络架构搜索中硬件在环延迟测量瓶颈及预测不准确问题，提出基于图注意力网络的HiFi-LLP延迟预测器，增加置信度度量，性能优于现有预测器，并构建混合NAS框架，实现加速且保持竞争力。

链接：https://arxiv.org/abs/2607.11746

机构：Technical University of Munich（慕尼黑工业大学）； BMW Group（宝马集团）； University of Bonn（波恩大学）

作者：Shambhavi Balamuthu Sampath, Behzad Shomali, Nael Fasfous, Moritz Thoma, Judeson Anthony Fernando, Lukas Frickenstein, Pierpaolo Mori, Manoj Rohit Vemparala, Alexander Frickenstein, Walter Stechele

AI 总结：研究利用自生成训练数据的顺序编码突破模型压缩极限，教师模型从学生分布选样本，学生代码记录选择，代码长度与参数和熵无关且更短，还能揭示现象、给出泛化保证、隔离可学习信息。

链接：https://arxiv.org/abs/2607.11883

机构：New York University（纽约大学）； Carnegie Mellon University（卡内基梅隆大学）

作者：Shikai Qiu, Marc Finzi, Yujia Zheng, Kun Zhang, Andrew Gordon Wilson

AI 总结：研究局部MoE推理在受限内存机器上的可行性，提出MawForge方法，通过磁盘存储模型、按需实例化专家张量等实现，发现其可作为有界执行机制，但非缓存最大化策略，性能受多种因素影响。

链接：https://arxiv.org/abs/2607.09686

机构：Holocron Security, Inc（全息安全公司）

作者：Craig Opie

AI 总结：研究旨在检测大语言模型是否经蒸馏而来，提出基于参考的成员推理蒸馏检测方法，通过比较模型与不同候选教师输出的优先对齐程度识别教师及蒸馏证据，能处理未知管道，扩展到开放世界设置，应用于当代模型获潜在蒸馏关系新证据。

链接：https://arxiv.org/abs/2607.09692

机构：University of California, Berkeley（加利福尼亚大学伯克利分校）； University of Southern California（南加利福尼亚大学）； OpenAI

作者：Rajat Rawat, Sizhe Chen, Akshay Anand, Michael Duan, Bob Rotsted, Sewon Min

AI 总结：针对机器学习训练资源供需差距，提出基于市场的配额市场机制，通过允许用户表达工作负载价值及动态定价，实现帕累托效率和最大最小公平性，促进符合组织优先级的资源分配。

链接：https://arxiv.org/abs/2607.09802

机构：Google（谷歌）； Drexel University（德雷塞尔大学）

作者：Balasubramanian Sivan, Renato Paes Leme, Mihai Tiuca, Ian McFarlane, Vasilis Gkatzelis, Nehal Mehta, Soheil Hassas Yeganeh, Vahab Mirrokni, Amin Vahdat

AI 总结：度假租赁市场供应端失衡，长尾房源难获有效服务。本文提出无需训练的大语言模型候选生成管道，用静态房产元数据补充IBKNN，经联合融合等策略提升效果，扩大候选覆盖范围，缩小模型召回差距，有效服务长尾房源。

链接：https://arxiv.org/abs/2607.09877

机构：Expedia Group（亿客行集团）

作者：Syed Mohammed Arshad Zaidi, Eric Rincon, Shayan Hassantabar

AI 总结：研究针对生成式检索中约束解码的瓶颈问题，提出FlashTrie方法。通过优化GPU上的约束束搜索，采用整数感知简洁trie布局和协作CUDA内核等技术，显著降低解码延迟、提高吞吐量，在实验中取得良好效果并提升了收入。

链接：https://arxiv.org/abs/2607.10044

机构：Microsoft（微软）； Nvidia（英伟达）

作者：Dakshitha Anandakumar, Anurag Mukkara, Wenxiang Hu, Jiusheng Chen, M Akash Kumar, Ting Ye, Qiang Lou, Jian Jiao

AI 总结：研究大语言模型PTQ在低精度下降问题，提出RDQ框架，通过级联误差补偿，在多种测试架构上取得最优结果，输出标准量化且零运行时开销。

链接：https://arxiv.org/abs/2607.10137

作者：Prateek Singh

AI 总结：研究跨模型价值比较中的两个混淆因素，提出无规则价值困境等方法分离并校正响应确定性，还发现访问约束影响模型价值概况，贡献了确定性校正分解，识别出部署约束是独特价值混淆因素。

链接：https://arxiv.org/abs/2607.10202

机构：KITECH（韩国产业技术评价院）； National Research Council of Science and Technology (NST)（韩国国家科学技术研究院）

作者：Hong-In Won, Jinseok Jang, Hyoseop Kim

AI 总结：研究可解释机器学习应用于心理健康结果时构建重叠问题，通过弹性网络管道及多组数据验证，发现特质焦虑和健康满意度占主导，经残差化实验揭示机制，指出残差化协议可用于相关XAI研究避免误判。

链接：https://arxiv.org/abs/2607.10633

机构：Sharif University of Technology（伊朗德黑兰谢里夫理工大学）； University of Southern California（美国南加州大学）

作者：Alireza Dehghan, Negin Ashrafi

AI 总结：研究大语言模型中参数重要性，提出权重调整梯度（WAG）方法，通过捕捉权重与梯度相互作用识别关键参数，揭示新的相互作用关系，展示其在多应用中的效用，为模型分析、调试和控制提供统一方法。

链接：https://arxiv.org/abs/2607.10803

机构：University of South Florida（南佛罗里达大学）； Brandeis University（布兰迪斯大学）

作者：Shrestha Datta, Hongfu Liu, Anshuman Chhabra

AI 总结：研究针对合成数据用于活动决策支持时决策不一致问题，提出策略模拟保真度标准、PolicySynth框架及三轴报告标准，通过实验对比，该框架在电信和银行语料库上展现高SSF及稳定性，可靠支持决策筛选，投资回报率虽需校正但表现良好。

链接：https://arxiv.org/abs/2607.11269

作者：Tung Dang, The Hung Phung, Son Lam Nguyen, Tu Nguyen

AI 总结：研究针对微调语言模型推理时安全对齐脆弱的问题，提出HyperSafe框架，通过生成特定模型的安全侧网络，利用层激活指纹和超网络映射参数，实现提示级安全分类，有效降低有害响应率，保持任务准确率。

链接：https://arxiv.org/abs/2607.11475

机构：King Abdullah University of Science and Technology（阿卜杜拉国王科技大学）

作者：Aznaur Aliev, Carlos Hinojosa, Abdelrahman Eldesokey, Bang An, Bernard Ghanem, Yibo Yang

AI 总结：研究选择性状态空间模型中状态使用，给出精确测量工具，能计算丢弃模式子集的输出误差。通过该工具发现模型依输入重分配状态空间，输入调度模式剪枝在各规模表现优，揭示了模型状态使用规律及有效剪枝方法。

链接：https://arxiv.org/abs/2607.11796

机构：Texas A&M University（德克萨斯农工大学）

作者：Raktim Bhattacharya

AI 总结：针对联邦学习中动态特征漂移问题，提出FedCausal-Dyn框架，通过因果域特征分离和因果特征引导协作正则化实现可靠动态原型聚合，在三个联邦域泛化基准实验中取得最优性能，为该场景下的联邦学习提供解决方案。

链接：https://arxiv.org/abs/2607.09695

机构：Mindlab（思维实验室）

作者：Kaijie Chen, Alex Johnson, Maria Garcia, Wei Zhang, Daniel Kim

AI 总结：研究对Turbo-Quant和SpectralQuant KV缓存压缩进行系统比较，用统计验证方法评估非支配方案，发现基于特征基的方法在重尾数据上因协方差不稳定失败，在结构化模式下表现出色，有效语义维度适应校准预算。

链接：https://arxiv.org/abs/2607.09683

机构：Advanced Micro Devices, Inc.（超威半导体公司）

作者：Paolo D'Alberto, Ashish Siarasao, Elliott Delaye, Rajeev Patwari

AI 总结：针对EHR数据多模态预测中现有方法缺乏可解释性的问题，提出显式多模态路由框架，通过构建不同模态路由及引入推理时路由掩码实现可解释、稳健且可审计的推理，经实验评估揭示了临床状况组模态依赖差异。

链接：https://arxiv.org/abs/2607.09982

机构：Southern Methodist University（南卫理公会大学）

作者：Nikkie Hooman, Zhongjie Wu, Eric C. Larson, Mehak Gupta

AI 总结：研究针对隐式神经表示的预测误差问题，提出轻量级方法，将基于回归的INR训练转为分类任务，通过离散连续目标为bins实现灵活分布建模，经分析权衡表明该方法能通过不确定性估计获高重建质量和误差感知。

链接：https://arxiv.org/abs/2607.10068

机构：Vanderbilt University（范德堡大学）； University of Arizona（亚利桑那大学）

作者：Zhimin Li, Jake D. Balla, Joshua A. Levine

AI 总结：研究多源检索增强生成中信息冲突问题，提出ERAG框架，通过将检索块转换为概率证据、用轻量级评估器和融合规则处理不确定性，实验表明该框架在标准问答有竞争力，在冲突情况下行为改善，是可信信息处理实用机制。

链接：https://arxiv.org/abs/2607.10491

机构：School of Computing, Wichita State University（威奇托州立大学计算学院）； Department of Computer Science, American International University-Bangladesh（美国国际孟加拉大学计算机科学系）

作者：S M Asif Hossain, Ruksat Khan Shayoni, M. F. Mridha

AI 总结：研究分布偏移对图神经网络校准的影响，给出校准的闭式理论表征，据此提出无源无标签校准方法STAC，实验证明其在校准上有改进，但无目标标签时可靠校准仍具挑战。

链接：https://arxiv.org/abs/2607.10804

机构：Computer Science and Applied Mathematics Laboratory (LIMA), Faculty of Science and Technology, Chadli Bendjedid University（计算机科学与应用数学实验室（LIMA），科学与技术学院，查德利·本杰迪德大学）

作者：Abderaouf Bahi

AI 总结：研究量化大语言模型的可靠性缩放定律，通过对其不确定性、校准和鲁棒性进行全面评估，刻画可靠性与模型比特总数的关系，发现4比特量化模型有可靠性峰值，且量化增强了模型对自然输入扰动的鲁棒性。

链接：https://arxiv.org/abs/2607.10855

机构：Technical University of Munich（慕尼黑工业大学）； Munich Data Science Institute（慕尼黑数据科学研究所）； Pruna AI（Pruna人工智能公司）

作者：Sirine Ayadi, Sándor Daróczi, Stephan Günnemann, Bertrand Charpentier

AI 总结：研究RLVR中奖励套件自然误报问题，通过预注册双臂因果对比实验，用GRPO在相同任务等条件下对比原始与强化测试奖励，发现平均留出效应有界，奖励误报与静态审计相关，还对评判者自身进行测试，表明廉价审计可定位暴露，强化奖励能消除测量膨胀。

链接：https://arxiv.org/abs/2607.11022

机构：Beijing Jiaotong University（北京交通大学）

作者：Chuyifei Zhang

AI 总结：研究如何利用证据神经网络在测试时通过邻居融合优化分子性质预测，提出PG-EVIKAL方法，学习性质距离度量重排邻居，经实验验证该方法能降低RMSE、改善校准，还能在连续测定场景中优化预测，证明证据不确定性分解可用于测试时分子性质预测优化。

链接：https://arxiv.org/abs/2607.11091

机构：Department of Chemical Engineering, University of Michigan（美国密歇根大学化学工程系）； Michigan Institute for Data & AI in Society, University of Michigan（美国密歇根大学数据与社会人工智能研究所）； Department of Statistics, University of Michigan（美国密歇根大学统计系）

作者：Cameron Gruich, Weichi Yao, Yixin Wang, Bryan Goldsmith

AI 总结：研究针对概率分类器事后校准方法，通过条件分层稳健性分析，比较TEMP和ISO在四个条件下的表现，评估四个假设组，发现数据集中稳健性因条件和指标而异，未提及外部可迁移性。

链接：https://arxiv.org/abs/2607.11542

机构：Fergana State Technical University（费尔干纳国立技术大学）

作者：Gurdeep Singh Virdee

AI 总结：研究针对黑箱条件分位数预测器，开发无分布且基于博弈论的测试框架，通过形式化校准概念、识别可实现功效的替代方案集，得出有限时间检测保证，能在特征层面解释证据过程，实证发现流行预测器存在校准错误。

链接：https://arxiv.org/abs/2607.11653

机构：Julius-Maximilians-Universität Würzburg（维尔茨堡大学朱利叶斯 - 马克西米利安斯大学）； Zaragoza Logistics Center（萨拉戈萨物流中心）； INSEAD（欧洲工商管理学院）

作者：Ivane Antonov, Sohom Mukherjee, Richard Pibernik, Yo Joong Choe

AI 总结：研究因果发现算法，指出常见流程中第一步的忠实性假设在自然系统中常被违反。提出干预即时忠实性假设，表明其足以用硬干预非参数识别因果结构，将干预作为因果结构信息主要载体，还针对干预范围有限情况指定等价类。

链接：https://arxiv.org/abs/2607.11816

机构：Thayer School of Engineering Dartmouth College（达特茅斯学院塞耶工程学院）； Broad Institute of MIT and Harvard（麻省理工学院和哈佛大学布罗德研究所）； Massachusetts Institute of Technology（麻省理工学院）

作者：Bijan Mazaheri, Jiaqi Zhang, Caroline Uhler

AI 总结：该综述针对知识图谱技术流程中基于GNN方法缺乏系统综述的问题，提出两级分类框架，分析GNN技术优势，详细回顾相关模型并总结优缺点，还讨论挑战与未来研究方向。

链接：https://arxiv.org/abs/2607.09666

机构：School of Safety Engineering, China University of Mining and Technology（中国矿业大学安全工程学院）； School of Computer Science and Technology, China University of Mining and Technology（中国矿业大学计算机科学与技术学院）； State Key Laboratory of Coal Mine Disaster Prevention and Control, China University of Mining and Technology（中国矿业大学煤矿灾害防治国家重点实验室）； Department of Computer Science, University of Illinois at Chicago（伊利诺伊大学芝加哥分校计算机科学系）

作者：Chengcheng Sun, Jiayun Tian, Cheng Zhai, Zhixiao Wang, Yajie Song, Xiaobin Rui, Jian Zhang, Philip S. Yu

AI 总结：针对大语言模型引导的进化程序搜索的瓶颈，提出GAE框架，通过图神经网络、强化学习优化的元控制器和在线微调循环解决问题，在复杂非线性振荡器系统符号回归任务中有效发现物理方程，性能达最优。

链接：https://arxiv.org/abs/2607.10127

作者：Xuanzhou Chen, Taoli Cheng

AI 总结：研究反洗钱图中GNN处理密集邻域问题，提出SALT-GNN轻量级架构，在消息传递层融合度感知统计聚合与注意力。实验表明其在减少参数同时提升密集上下文F1分数，增益与注意力风格无关。

链接：https://arxiv.org/abs/2607.10131

作者：Lidia Losavio, Francesco Sovrano, Dario Fenoglio, Martin Gjoreski, Marc Langheinrich

AI 总结：该研究针对计算群论中判断有限群子群关系的基本问题，提出用连体图神经网络，结合有限群的凯莱图表示与代数特征来学习子群关系，实验证明此方法有效，在独立测试集上准确率达95.9% 。

链接：https://arxiv.org/abs/2607.11140

作者：Tal Weissblat

AI 总结：研究如何让表格基础模型有效捕获图结构信息，提出GTAlign框架，先预训练图编码器捕获图表示，通过社区引导持续预训练弥合差距，应用于目标域推理，实验表明该框架在节点和图分类上显著优于基线，提供简单无文本的图基础模型。

链接：https://arxiv.org/abs/2607.11374

机构：Nankai University（南开大学）； Beihang University（北京航空航天大学）； Tsinghua University（清华大学）

作者：Chunyu Hu, Tianyin Liao, Ge Lan, Xingxuan Zhang, Jianxin Li, Peng Cui, Ziwei Zhang

AI 总结：该研究针对传统GNN问题，引入约束双视图框架，利用锚定网络捕获特征，提出CSAG-GNN及循环交替优化策略，在节点预测中实现结构特征对齐，实证结果显示相比基线有性能提升和结构稳健性。

链接：https://arxiv.org/abs/2607.11577

机构：College of Artificial Intelligence, Shaoxing Institute of Technology（绍兴文理学院人工智能学院）； School of Mathematics and Computational Science, Xiangtan University（湘潭大学数学与计算科学学院）

作者：Chengcheng Yan, Qingsong Wang

AI 总结：针对车辆和航空航天工程等领域流体动力学模拟成本高的问题，提出多尺度特征增强图神经网络（ME-GNN），通过两步消息传递、集成注意力U-Net等方法，在三个基准数据集上取得了先进的流体动力学预测结果。

链接：https://arxiv.org/abs/2607.11672

作者：Li Xiao, Tianyu Li, Yiye Zou, Mingjie Zhang, Xiaogangd Deng

AI 总结：研究针对状态空间模型压缩历史信息能力有限的问题，提出延迟状态空间模型DSSMs，通过显式延迟状态反馈增强离散SSM循环，推导传递函数在频域计算内核，解决相关工具难题，在延迟检索任务上有显著改进，性能优于S4D。

链接：https://arxiv.org/abs/2607.10244

机构：College of Control Science and Engineering, Zhejiang University（浙江大学控制科学与工程学院）； Department of Mathematics, National University of Singapore（新加坡国立大学数学系）； School of Mathematical Sciences, Zhejiang University（浙江大学数学科学学院）

作者：Yixiao Qian, Song Chen, Jiaxu Liu, Shengze Cai, Chao Xu

AI 总结：针对辐射传输模拟中M1方法简化系统不封闭问题，提出双曲神经封闭模型。通过两个神经网络对雅可比矩阵参数化确保实特征值，经数值积分重建封闭模型。该模型提高了封闭及解的精度，在相关模拟中保持稳定。

链接：https://arxiv.org/abs/2607.10364

机构：Purdue University（普渡大学）； University of Notre Dame（圣母大学）

作者：Bongseok Kim, Jiahao Zhang, Johannes Krotz, Dinshaw Balsara, Ryan McClarren, Guang Lin

AI 总结：针对可穿戴设备上心电图分类计算成本高的问题，提出轻量级混合模型LSTrans，它结合一维卷积主干、Transformer编码器及知识蒸馏，在多基准数据集实验中实现诊断灵敏度与资源效率的平衡，减少内存占用和训练延迟。

链接：https://arxiv.org/abs/2607.10784

机构：School of Mechano-Electronic Engineering Xidian University Xi'an, China（机械电子工程学院 西安电子科技大学 西安中国）； School of Cyber Engineering Xidian University Xi'an, China（网络工程学院 西安电子科技大学 西安中国）； DCST, BNRist, RIIT, Institute of Internet Industry Tsinghua University Beijing, China（DCST、BNRist、RIIT、互联网产业研究院 清华大学 北京中国）

作者：Yi Zhao, Jiajun Gao, Chenyang Xu, Yuxi Zhou, Hao Wang

AI 总结：研究如何通过链路自适应算法选择MCS类型，提出联合汤普森采样算法，利用MCS成功概率有序的特点，采用多元有序贝塔分布作为先验，解决现有算法在特定场景的不足，实现各场景下有竞争力的吞吐量和稳健性能。

链接：https://arxiv.org/abs/2607.11075

机构：IIT Hyderabad（印度海得拉尔理工学院）

作者：Vignatha Vinjam, Manjunath Kolavennu, Myna Vajha, Karthik Periyapattana Narayanaprasad

AI 总结：研究动态图持续学习中现有方法未充分利用时间信息的问题。提出CA-DGCL框架，先凝聚历史图快照，再构建节点链经塔克分解获取稳定节点特征以重放过去信息，引入改进遗忘度量。实验证明该方法在遗忘抑制和准确率上表现出色。

链接：https://arxiv.org/abs/2607.11112

机构：College of Computer and Information Science, Southwest University（西南大学计算机与信息科学学院）

作者：Tingxu Yan Ye Yuan

AI 总结：探讨机器学习中真实数据集非客观真理而是人为构建，主张提高‘情境可靠性’，关注其构建能提升透明度、问责制等，利于阐明模型优缺点及更好运用数据集与模型。

链接：https://arxiv.org/abs/2607.09668

作者：Charlotte Högberg, Ericka Johnson, Kiri L. Wagstaff

AI 总结：研究编码智能体编辑代码时实际所需上下文，通过固定定位、改变代码表示并评分，发现代码本身自然语言摘要作用小，周围上下文无关紧要，压缩上下文效果好，还发现API推理的噪声底限，且发布相关工具。

链接：https://arxiv.org/abs/2607.09691

机构：Integrallis Software（Integrallis软件公司）

作者：Brian Sam-Bodden

AI 总结：研究针对点击率预测模型早期训练崩溃问题，通过大规模工业数据集分析，发现降低学习率效果不佳，而控制特征稀疏性如去除高度稀疏特征、聚合罕见特征值可稳定训练，提升离线和在线性能。

链接：https://arxiv.org/abs/2607.09696

机构：Huawei Türkiye R&D Center（华为土耳其研发中心）

作者：Ergun Biçici, Erkan Çetinyamaç

AI 总结：研究动态交易图中稀疏环欺诈检测问题，用合成交易模拟器，将每日交易图聚合到滚动窗口并以多种方式表示。比较GRU基线和量子启发式CML，结果表明拓扑作上下文层有用，CML是检测此类欺诈有前途的候选模型。

链接：https://arxiv.org/abs/2607.09704

机构：Infleqtion Australia（澳大利亚Infleqtion公司）； SoftBank Corp.（软银集团）

作者：Behnam Tonekaboni, Hiroshi Yamauchi

AI 总结：针对表格分类中离散规则结构与欧氏空间DNN几何不匹配问题，提出HDE-Net，通过抽象特征为LDN并嵌入双曲空间，引入软决策路由机制和容量分配算法，在分类基准上取得最佳平均排名，优于其他方法。

链接：https://arxiv.org/abs/2607.09710

机构：Newcastle University（纽卡斯尔大学）

作者：Tian Li, Lucy Robinson, Varun Ojha, Huizhi Liang

AI 总结：研究智能体能否从自身运行学可复用技能，引入EvoClawBench基准测试，涵盖多任务多子问题并支持多运行时。实验对比不同方式，发现直接基线性能依赖运行时，自我创作技能效果各异，表明学习可复用技能有选择性且成本敏感。

链接：https://arxiv.org/abs/2607.09711

机构：Shanghai Jiao Tong University（上海交通大学）； Zhejiang University（浙江大学）； Hithink Research（同花顺研究院）

作者：Zhiyuan Peng, Xin Yin, Chenhao Ying, Zhe Cui, Zixiang Ding, Zhenhua Liu, Jiang Wu, Yuan Luo

AI 总结：研究针对财务控制测试依赖ERP数据但直接复制有风险的问题，提出SEQ-FCT框架，结合多种技术，用合成数据集评估，与多种基线对比，在对账、欺诈触发召回率、控制失败等方面取得较好结果，表明综合评估更可靠。

链接：https://arxiv.org/abs/2607.09712

作者：Anitha Samudrala

AI 总结：研究金融应用中多模态事件序列预测建模，提出预训练基础变压器模型统一多源事件成序列，经下一个事件预测学习通用表示，结合现有特征训练轻量级神经模型用于多下游任务，优于传统模型并减少开发开销且已应用取得业务改进。

链接：https://arxiv.org/abs/2607.09955

机构：Sber（俄罗斯储蓄银行）； Sber AI Lab（俄罗斯储蓄银行人工智能实验室）

作者：Nikita Rusakov, Vladislav Meshkov, Konstantin Zorin, Gleb Zaripov, Alexander Uglov, Alexey Vasilev, Anton Klenitskiy

AI 总结：针对视频语言模型物理合理性推理不可靠的问题，提出免训练的PhysMRV框架，通过将训练视频转化为分层内存库，利用结构化物理证据指导VLM验证物理合理性，在多基准测试中取得一致改进，有效增强该推理能力。

链接：https://arxiv.org/abs/2607.10190

作者：Wenyuan Wang, Lianyu Hu, Hao Wang, Yang Liu

AI 总结：研究对深度学习架构进行比较评估，用于农业气象参数多变量预测，利用希腊约阿尼纳的观测数据，对比多种模型在24小时次日和168小时提前一周预测任务中的表现，发现混合模型在短期预测中更优且参数更少。

链接：https://arxiv.org/abs/2607.10208

机构：Lodz Universtiy of Technlogy（罗兹工业大学）； MicroComputer Systems Laboratory（微计算机系统实验室）； University of Ioannina（约阿尼纳大学）

作者：Piotr Sikora, Sotirios Kontogiannis

AI 总结：为解决全模态理解不足问题，提出AVDC数据集及AVDC-QA-CoT数据集，利用现成模型标注视频，采用两阶段训练范式，在多下游任务实验中取得显著性能提升，推动全模态感知发展。

链接：https://arxiv.org/abs/2607.10299

机构：SAI, Shanghai Jiao Tong University（上海交通大学 上海人工智能研究院）； Zhejiang University（浙江大学）； Shanghai AI Lab（上海人工智能实验室）

作者：Kaiying Yan, Luoyi Sun, Xiao Zhou, Weidi Xie

AI 总结：研究针对核结构材料小尺寸与标准尺寸试样夏比冲击性能相关性问题，提出基于机器学习的框架，通过特定映射和模型提取相关值，经验证其相关性优于传统方法，且推理无需全尺寸数据，适用于多种相关应用。

链接：https://arxiv.org/abs/2607.10412

作者：Yugandhar Kasala Sreenivasulu, Isshu Lee, John W. Merickel, Fei Xu, Yalei Tang, Joshua E. Rittenhouse, Aleksandar Vakanski, Rongjie Song

AI 总结：研究生存模型中行政审查的陷阱，通过刻画行政截断泄漏发生的情况，将其与其他情况区分，提出检测方法，经模拟和真实队列验证，为生存预测提出数据集应提供足够潜在随访的设计原则。

链接：https://arxiv.org/abs/2607.10466

机构：Center for Data Science, New York University（纽约大学数据科学中心）； University of Chicago（芝加哥大学）； NYU Grossman School of Medicine（纽约大学格罗斯曼医学院）； Ataraxis AI（Ataraxis人工智能公司）

作者：Yanqi Xu, Hui Dai, Carlos Fernandez-Granda, Krzysztof J. Geras, Yiqiu Shen

AI 总结：研究旨在从噪声数据发现偏微分方程，提出LLM-PDESR框架，结合大语言模型假设生成与数学评估环境，利用五次样条和子域加权残差减轻噪声影响，经帕累托驱动反馈环优化方程，实验表明其在多方面显著优于现有方法。

链接：https://arxiv.org/abs/2607.10546

机构：College of Computer Science and Technology, Jilin University（吉林大学计算机科学与技术学院）； School of Big Data and Artificial Intelligence, Guangdong University of Finance and Economics（广东财经大学大数据与人工智能学院）； Key Laboratory of Symbolic Computation and Knowledge Engineering of Ministry of Education, Jilin University（吉林大学符号计算与知识工程教育部重点实验室）； School of Computer Science, Zhuhai College of Science and Technology（珠海科技学院计算机科学学院）； Department of Mechanical Engineering, National University of Singapore（新加坡国立大学机械工程系）

作者：Jinyang Du, Hao Ma, Xiaohu Shi, Bo Yang, Yanchun Liang, Heow Pueh Lee, Chunguo Wu

AI 总结：研究开放权重语言模型谱系验证问题，提出modelDNA工具，通过HTTP读取指纹识别模型，与参考数据库比较并返回判定类别，实现高AUROC和零误报，还能进行合并分解，恢复混合权重，公开相关数据可离线重现。

链接：https://arxiv.org/abs/2607.10617

作者：Muhammad Awais Bin Adil, Saad Aamir

AI 总结：研究ADMET模型中支架分割掩盖结构前沿失败问题，引入无标签结构前沿分割并评估，对比多种分割方法，发现前沿分割会增大误差，多视图前沿风险外推法等测试结果不明确，强调分割构建和标签来源对评估的重要性。

链接：https://arxiv.org/abs/2607.10729

机构：Ma Yinchu School of Economics, Tianjin University（天津大学马寅初经济学院）； Marine College, Shandong University（山东大学海洋学院）

作者：Jiacheng Zheng, Chang Guo, Zixuan Wang, Xinyu Liu

AI 总结：研究针对数值积分中贝叶斯求积法因基于平稳协方差函数导致对非平稳被积函数模型错误指定的问题，提出通过积分域树状划分成局部平稳模型的方法，经GP条件层次结构重组估计，提升了非平稳被积函数积分效果。

链接：https://arxiv.org/abs/2607.10793

机构：Tübingen AI Center（图宾根人工智能中心）； University of Tübingen（图宾根大学）； School of Engineering Sciences（工程科学学院）； Lappeenranta–Lahti University of Technology（拉佩伦塔-拉赫蒂技术大学）

作者：Tim Weiland, Toni Karvonen, Philipp Hennig

AI 总结：研究耦合目标\(J_K^{WOR}\)，现有样本重用方法在此有偏差。提出用秩条件霍维茨 - 汤普森估计为\(J_K^{WOR}\)子集总量构建无偏估计器，含动态规划简化计算，还介绍了相关性质及未解决问题。

链接：https://arxiv.org/abs/2607.11146

机构：Independent Researcher（独立研究者）

作者：Melveena Jolly, Midhun Xavier

AI 总结：研究针对大语言模型时间事实更新问题，基于因果追踪发现其内部计算支持新旧答案区分，进而引入PRISM Edit，通过优化单一多义词表示及利用固有调制路径，在新基准上评估，相比基线提升了时间一致性等指标且速度更快。

链接：https://arxiv.org/abs/2607.11327

机构：Tsinghua University（清华大学）； ByteDance（字节跳动）

作者：Chen Huang, Qi Zheng, Ruiqin Zheng, Long Zeng, Yuantong Xu

AI 总结：针对现实世界多数数据是嵌入高维欧几里得空间流形的情况，提出基于内在几何的生成对抗网络IG-GAN用于空气动力学数据生成。其生成器用贝塞尔曲面构建流形，判别器为RBF-D。实验显示该方法相比基线能显著降低预测均方误差。

链接：https://arxiv.org/abs/2607.11497

机构：School of Information Science and Engineering, Hebei University of Science and Technology（信息科学与工程学院，河北科技大学）

作者：Ying Yan, Liwei Hu, Xiaoming Zhang

AI 总结：研究固定预算最佳臂识别问题，对比算法与静态预言机，证明对于任意\(K\geq3\)及特定奖励，任何算法至少在一个实例中的错误衰减率至多为静态预言机的\((1 + \frac{\log(K)}{8})^{-1}\)倍，回答了相关开放问题。

链接：https://arxiv.org/abs/2607.11635

机构：Yale University（耶鲁大学）

作者：Motti Goldberger

AI 总结：研究针对人工智能辅助和数据转换工作流程中证据追溯及防篡改问题，提出轻量级Python库AuditWeave，它能将工作流程步骤记录到哈希链接账本，通过链验证检测篡改，评估了其记录开销等性能，保证了事件完整性。

链接：https://arxiv.org/abs/2607.09682

作者：Vimal Nakrani

AI 总结：研究强迫物理系统中潜在响应规律，提出FLARE方法，通过学习紧凑响应坐标等，能恢复强迫动力学并预测高维响应，还可将方程发现扩展到复杂观测系统，为可解释建模和预测提供途径。

链接：https://arxiv.org/abs/2607.09801

机构：Beijing University of Technology（北京工业大学）； State Key Laboratory of Bridge Safety and Resilience（桥梁安全与韧性国家重点实验室）

作者：Yi Zhu, Su Chen, Xiaojun Li, Xiuli Du

AI 总结：针对室内空间理解难题，本文提出基于图神经网络的学习框架，整合信号强度等数据构建图表示，通过 GNN 训练预测空间几何模式，如线性轨迹等，为室内定位等提供新方法。

链接：https://arxiv.org/abs/2607.10822

作者：Curtis Shull, Merrick Green, Roy Rucker

AI 总结：研究符号分支重复惩罚中规范依赖性和结构化输出损坏问题，发现其惩罚定义不明确且损坏输出，将惩罚应用于归一化对数概率可消除这些问题，给出了相关机制、测量及归一化变体。

链接：https://arxiv.org/abs/2607.09791

作者：Peter Hollows

AI 总结：研究长尾识别中仅再训练分类器能否减少尾部误差，提出BS - cRT方法，在多个数据集上实验，该方法持续提高少样本准确率，还分析了CBRM的失败模式，支持BS - cRT为实用基线，表明边界监督需考虑类频率。

链接：https://arxiv.org/abs/2607.09832

机构：CICATA Querétaro Instituto Politecnico Nacional（国立理工学院克雷塔罗中心研究与高级研究中心）； Facultad de Informática Universidad Autónoma de Querétaro（克雷塔罗自治大学信息学院）； Escuela Nacional de Estudios Superiores Universidad Nacional Autónoma de México（国立自治大学高等研究学校）

作者：Juan Terven, Diana Margarita Córdova Esparza, Julio Alejandro Romero Gonzalez, Edgar Arturo Chávez Urbiola, Francisco Javier Willars Rodriguez, Juan Bautista Hurtado Ramos, Alfonso Ramirez Pedraza

AI 总结：研究合作博弈中归因问题，受最小核心启发引入一类非线性公理归因方法，该方法产生的贡献向量是特定最小化问题的唯一最优解，实验表明其在包含AUC指标上比仅放宽效率公理的夏普利值变体更有效。

链接：https://arxiv.org/abs/2607.09869

机构：National University of Singapore（新加坡国立大学）； University of Waterloo（滑铁卢大学）； Vector Institute（向量研究所）

作者：Weida Li, Zhuanghua Liu, Yaoliang Yu, Bryan Kian Hsiang Low

AI 总结：研究针对现有大环肽结构和性质建模方法的局限，提出全原子深度学习模型Vilya-1，可跨化学结构采样及预测属性，在异构数据集训练，提高几何精度且覆盖小分子，支持生成应用，为大环治疗药物开发加速。

链接：https://arxiv.org/abs/2607.09998

作者：Vilya Research, :, Pascal Sturmfels, Milad Salem, Naozumi Hiranuma, Stephen Rettie, Xiaoliang Pan, Benjamin D. Sellers, Adam P. Moyer, Patrick J. Salveson, Ivan Anishchanka

AI 总结：针对大规模表格学习，提出TabLoRA方法，通过共享通用主干和引入特定预测器的低秩适配，实现参数高效的神经集成学习，在相同资源约束下平衡了预测性能与实际效率，提升了神经集成学习的可行性。

链接：https://arxiv.org/abs/2607.10077

机构：School of Mathematical Sciences, Soochow University（苏州大学数学科学学院）； Digital innovation research center, Duke Kunshan University（昆山杜克大学数字创新研究中心）

作者：Jiaqi Luo, Shixin Xu

AI 总结：研究改进语言模型中位置编码的方法，提出可学习RoPE频率的LeRoPE方法，通过训练不同规模语言模型验证，该方法在各规模下均优于RoPE和部分RoPE，提升了语言建模性能。

链接：https://arxiv.org/abs/2607.10134

机构：UC San Diego（加利福尼亚大学圣地亚哥分校）

作者：Petros Karypis, Sean O'Brien, Shreyas Kadekodi, Rui Zhu, Julian McAuley

AI 总结：研究大语言模型为Unity游戏场景编写代码，去除迭代修复循环进行单遍生成评估，通过对多个模型、模式等生成的代码分析编译器错误，发现瓶颈是缺少引擎特定知识，按需求对目标模式排序展示单遍生成断点。

链接：https://arxiv.org/abs/2607.10187

机构：Chalmers University of Technology（查尔姆斯理工大学）； University of Gothenburg（哥德堡大学）

作者：Hugh Xuechen Liu, Kıvanç Tatar

AI 总结：研究针对物理信息神经网络（PINN），提出微分神经切线核（DNTK）理论框架，通过NTK分析PINN，建立了无限宽度DNTK的正性，为训练PINN的梯度型算法分析奠定基础。

链接：https://arxiv.org/abs/2607.10200

机构：The Chinese University of Hong Kong（香港中文大学）

作者：Bangti Jin, Longjun Wu

AI 总结：研究探讨潜在世界模型中深度在组合时的情况，用浅度惩罚ρ测试九个深度思维控制任务，发现三种机制，揭示反转机制由训练产生，其与观察/动作维度等相关，还指出任务机制受多种因素影响及相关注意事项。

链接：https://arxiv.org/abs/2607.10203

机构：New York University（纽约大学）

作者：Achyuthan Sivasankar

AI 总结：研究针对上下文工程中以令牌计量信息的问题，基于特定机制开发按新颖性分配的缓存作为工作记忆组件，依任务对过去的依赖组织上下文，通过实验验证该方法在不同场景下的有效性及优势，可实现可审计的上下文管理。

链接：https://arxiv.org/abs/2607.10441

作者：Siddharth Pal, Viktoria Rojkova

AI 总结：研究受限玻尔兹曼机在分布外输入下的脆弱性，通过分析可见-可见相互作用\(J\)的谱，提出用随机辅助曝光使相互作用有效秩坍缩，能在保持MNIST准确率时拒绝结构化OOD图像数据集。

链接：https://arxiv.org/abs/2607.10506

机构：University of Virginia（弗吉尼亚大学）

作者：Oshada Rathnayake, Nikhil Shukla

AI 总结：研究从算子视角理解自注意力，将令牌序列视为向量场，证明单头和多头注意力的性质及相关条件，通过实验发现不同规模和结构的训练Transformer呈现与理论一致的几何结构，提供连接自注意力与几何算子的形式及分析工具。

链接：https://arxiv.org/abs/2607.10677

机构：Zhejiang University（浙江大学）； Alibaba Cloud（阿里云）

作者：Binbin Lin, Wei Chen, Yalun Li, Wenxiao Wang, Jieping Ye, Xiaofei He

AI 总结：研究高维域在线比例采样问题，设计支持高效更新和采样的数据结构，在不同对手下证明数据结构深度界，应用于在线学习获高效无悔算法及次线性遗憾保证。

链接：https://arxiv.org/abs/2607.10963

作者：Amirmahdi Mirfakhar, Maria-Florina Balcan, Hedyeh Beyhaghi

AI 总结：研究针对因残疾等行动不便患者，提出基于事件的神经解码方法，用基于事件的门控循环单元生成稀疏通信模式，经有效训练方法和稀疏推理，在任务性能上超越经典脉冲神经网络，为设备上神经解码带来新机会。

链接：https://arxiv.org/abs/2607.11445

机构：TU Dresden（德累斯顿工业大学）

作者：Khaleelulla Khan Nazeer, Sirine Arfa, Matthias Jobst, Richard George, Christian Mayr

AI 总结：研究大语言模型作为评判者的偏差，提出偏差在隐藏状态有表示层面解释。通过七位评判者等实验发现，偏差输入沿特定子空间移动，操纵隐藏状态可控制评分，线性投影能预测评判失败，统一多方面内容。

链接：https://arxiv.org/abs/2607.11871

机构：AMAP, Alibaba Group（阿里巴巴集团AMAP实验室）； Mohamed bin Zayed University of Artificial Intelligence（穆罕默德·本·扎耶德人工智能大学）； University of Southern California（南加州大学）； University of Michigan, Ann Arbor（密歇根大学安娜堡分校）

作者：Zixiang Xu, Sixian Li, Huaxing Liu, Xiang Wang, Shuai Li, Zirui Song, Xiuying Chen

AI 总结：该研究提出理论框架解释Transformer语言模型归纳推理能力，研究广义归纳任务，证明其训练动态可限制在低维不变流形，刻画数据统计对学习竞争的影响，探讨初始化作用并展示坐标框架用途，向Transformer学习预测理论迈进。

链接：https://arxiv.org/abs/2607.11875

机构：ETH Zurich（苏黎世联邦理工学院）； Stanford（斯坦福大学）

作者：Tiberiu Musat, Tiago Pimentel, Nicholas Zucchet, Thomas Hofmann