机器学习学术速递[8.1]

点击阅读原文访问arxivdaily.com，涵盖CS|物理|数学|经济|统计|金融|生物|电气领域，更有搜索、收藏等功能！

cs.LG 方向，今日共计114篇

大模型相关(18篇)

【1】SimuRA: Towards General Goal-Oriented Agent via Simulative Reasoning Architecture with LLM-Based World Model
标题：SimuRA：通过具有基于LLM的世界模型的模拟推理架构迈向通用目标导向代理
链接：https://arxiv.org/abs/2507.23773

【2】villa-X: Enhancing Latent Action Modeling in Vision-Language-Action Models
标题：villa-X：增强视觉-语言-动作模型中的潜在动作建模
链接：https://arxiv.org/abs/2507.23682

【3】From LLMs to Edge: Parameter-Efficient Fine-Tuning on Edge Devices
标题：从LLM到边缘：边缘设备上的参数高效微调
链接：https://arxiv.org/abs/2507.23536

【4】Policy Learning from Large Vision-Language Model Feedback without Reward Modeling
标题：从大型视觉语言模型反馈中进行政策学习，无需奖励建模
链接：https://arxiv.org/abs/2507.23391

【5】MUST-RAG: MUSical Text Question Answering with Retrieval Augmented Generation
标题：MUST-RAG：具有检索增强生成的音乐文本问题解答
链接：https://arxiv.org/abs/2507.23334

【6】DynaSwarm: Dynamically Graph Structure Selection for LLM-based Multi-agent System
标题：DynaSwarm：基于LLM的多智能体系统的动态图结构选择
链接：https://arxiv.org/abs/2507.23261

【7】Evaluating LLMs' Multilingual Capabilities for Bengali: Benchmark Creation and Performance Analysis
标题：评估LLM的孟加拉语多语言能力：基准创建和性能分析
链接：https://arxiv.org/abs/2507.23248

【8】Enabling Few-Shot Alzheimer's Disease Diagnosis on Tabular Biomarker Data with LLMs
标题：利用LLM对表格生物标志物数据进行Few-Shot阿尔茨海默病诊断
链接：https://arxiv.org/abs/2507.23227

【9】Not Just What, But When: Integrating Irregular Intervals to LLM for Sequential Recommendation
标题：不仅仅是什么，而是何时：将不规则间隔整合到LLM中以进行顺序推荐
链接：https://arxiv.org/abs/2507.23209

【10】BAR Conjecture: the Feasibility of Inference Budget-Constrained LLM Services with Authenticity and Reasoning
标题：BAR猜想：具有真实性和推理的推理预算约束LLM服务的可行性
链接：https://arxiv.org/abs/2507.23170

【11】LENS: Learning Ensemble Confidence from Neural States for Multi-LLM Answer Integration
标题：LENS：从神经状态学习增强信心以实现多LLM答案集成
链接：https://arxiv.org/abs/2507.23167

【12】KLLM: Fast LLM Inference with K-Means Quantization
标题：KlLM：带K均值量化的快速LLM推理
链接：https://arxiv.org/abs/2507.23035

【13】LLM-Assisted Cheating Detection in Korean Language via Keystrokes
标题：LLM辅助击键检测韩语作弊
链接：https://arxiv.org/abs/2507.22956

【14】LLMs Between the Nodes: Community Discovery Beyond Vectors
标题：节点之间的LLM：超越载体的社区发现
链接：https://arxiv.org/abs/2507.22955

【15】Automated Label Placement on Maps via Large Language Models
标题：通过大型语言模型在地图上自动放置标签
链接：https://arxiv.org/abs/2507.22952

【16】ELMES: An Automated Framework for Evaluating Large Language Models in Educational Scenarios
标题：ELMES：评估教育场景中大型语言模型的自动化框架
链接：https://arxiv.org/abs/2507.22947

【17】Semantic Convergence: Investigating Shared Representations Across Scaled LLMs
标题：语义融合：调查跨规模LLM的共享表示
链接：https://arxiv.org/abs/2507.22918

【18】A Language Model-Driven Semi-Supervised Ensemble Framework for Illicit Market Detection Across Deep/Dark Web and Social Platforms
标题：语言模型驱动的半监督集成框架，用于跨深网/暗网和社交平台的非法市场检测
链接：https://arxiv.org/abs/2507.22912

Graph相关(图学习|图神经网络|图优化等)(4篇)

【1】Rule2Text: Natural Language Explanation of Logical Rules in Knowledge Graphs
标题：规则2文本：知识图中逻辑规则的自然语言解释
链接：https://arxiv.org/abs/2507.23740

【2】GraphRAG-R1: Graph Retrieval-Augmented Generation with Process-Constrained Reinforcement Learning
标题：GraphRAG-R1：基于过程约束强化学习的图检索增强生成
链接：https://arxiv.org/abs/2507.23581

【3】Scalable Generative Modeling of Weighted Graphs
标题：加权图的可扩展生成建模
链接：https://arxiv.org/abs/2507.23111

【4】Unifying Post-hoc Explanations of Knowledge Graph Completions
标题：知识图完备化的统一事后证明
链接：https://arxiv.org/abs/2507.22951

GAN|对抗|攻击|生成相关(4篇)

【1】XSpecMesh: Quality-Preserving Auto-Regressive Mesh Generation Acceleration via Multi-Head Speculative Decoding
标题：XSpecGrid：通过多头推测解码加速质量保护的自回归网格生成
链接：https://arxiv.org/abs/2507.23777

【2】DivControl: Knowledge Diversion for Controllable Image Generation
标题：DivControl：可控图像生成的知识转移
链接：https://arxiv.org/abs/2507.23620

【3】Zero-Shot Document Understanding using Pseudo Table of Contents-Guided Retrieval-Augmented Generation
标题：使用伪内容表引导检索增强生成的Zero-Shot文档理解
链接：https://arxiv.org/abs/2507.23217

【4】AI paradigm for solving differential equations: first-principles data generation and scale-dilation operator AI solver
标题：用于求解方程的人工智能范式：第一性原则数据生成和规模膨胀操作符人工智能求解器
链接：https://arxiv.org/abs/2507.23141

半/弱/无/有监督|不确定性|主动学习(6篇)

【1】Improving annotator selection in Active Learning using a mood and fatigue-aware Recommender System
标题：使用情绪和疲劳感知推荐系统改进主动学习中的注释者选择
链接：https://arxiv.org/abs/2507.23756

【2】Deep Learning-based Prediction of Clinical Trial Enrollment with Uncertainty Estimates
标题：基于深度学习的临床试验招募预测和不确定性估计
链接：https://arxiv.org/abs/2507.23607

【3】Incorporating structural uncertainty in causal decision making
标题：因果决策中的结构性不确定性
链接：https://arxiv.org/abs/2507.23495

【4】An Interpretable Data-Driven Unsupervised Approach for the Prevention of Forgotten Items
标题：一种可解释的数据驱动无监督方法来防止遗忘物品
链接：https://arxiv.org/abs/2507.23303

【5】Are Recommenders Self-Aware? Label-Free Recommendation Performance Estimation via Model Uncertainty
标题：推荐人有自知之明吗？通过模型不确定性的无标签推荐性能估计
链接：https://arxiv.org/abs/2507.23208

【6】DICOM De-Identification via Hybrid AI and Rule-Based Framework for Scalable, Uncertainty-Aware Redaction
标题：通过混合人工智能和基于规则的框架进行的可扩展、不确定性编辑的DICIC去识别
链接：https://arxiv.org/abs/2507.23736

迁移|Zero/Few/One-Shot|自适应(3篇)

【1】Anomalous Samples for Few-Shot Anomaly Detection
标题：用于Few-Shot异常检测的异常样本
链接：https://arxiv.org/abs/2507.23712

【2】AGA: An adaptive group alignment framework for structured medical cross-modal representation learning
标题：AGA：用于结构化医学跨模式表示学习的自适应群体对齐框架
链接：https://arxiv.org/abs/2507.23402

【3】Formal Bayesian Transfer Learning via the Total Risk Prior
标题：通过总风险先验的正式Bayesian转移学习
链接：https://arxiv.org/abs/2507.23768

强化学习(3篇)

【1】OptiGradTrust: Byzantine-Robust Federated Learning with Multi-Feature Gradient Analysis and Reinforcement Learning-Based Trust Weighting
标题：OptGradTrust：具有多特征梯度分析和基于强化学习的信任加权的拜占庭鲁棒联邦学习
链接：https://arxiv.org/abs/2507.23638

【2】Hierarchical Message-Passing Policies for Multi-Agent Reinforcement Learning
标题：多智能体强化学习的分层消息传递策略
链接：https://arxiv.org/abs/2507.23604

【3】Generalized Reinforcement Learning for Retriever-Specific Query Rewriter with Unstructured Real-World Documents
标题：具有非结构化现实世界文档的检索特定查询重写器的广义强化学习
链接：https://arxiv.org/abs/2507.23242

医学相关(4篇)

【1】SAMSA: Segment Anything Model Enhanced with Spectral Angles for Hyperspectral Interactive Medical Image Segmentation
标题：SAMAS：利用光谱角度增强的分割Anything模型，用于高光谱交互式医学图像分割
链接：https://arxiv.org/abs/2507.23673

【2】Explainable artificial intelligence model predicting the risk of all-cause mortality in patients with type 2 diabetes mellitus
标题：预测2型糖尿病患者全因死亡风险的可解释人工智能模型
链接：https://arxiv.org/abs/2507.23491

【3】Machine learning and machine learned prediction in chest X-ray images
标题：胸部X射线图像中的机器学习和机器学习预测
链接：https://arxiv.org/abs/2507.23455

【4】SigBERT: Combining Narrative Medical Reports and Rough Path Signature Theory for Survival Risk Estimation in Oncology
标题：SigBERT：结合叙述性医疗报告和粗糙路径签名理论进行肿瘤学生存风险估计
链接：https://arxiv.org/abs/2507.22941

自动驾驶|车辆|车道检测等(2篇)

【1】Vision-Language Fusion for Real-Time Autonomous Driving: Goal-Centered Cross-Attention of Camera, HD-Map, & Waypoints
标题：实时自动驾驶的视觉语言融合：摄像机、高清地图和航点的以目标为中心的交叉注意力
链接：https://arxiv.org/abs/2507.23064

【2】Early Goal-Guided Multi-Scale Fusion for Real-Time Vision-Language Driving
标题：用于实时视觉语言驱动的早期目标引导多尺度融合
链接：https://arxiv.org/abs/2507.23042

联邦学习|隐私保护|加密(3篇)

【1】FLOSS: Federated Learning with Opt-Out and Straggler Support
标题：FLOSS：有选择退出和落后者支持的联合学习
链接：https://arxiv.org/abs/2507.23115

【2】FedCVD++: Communication-Efficient Federated Learning for Cardiovascular Risk Prediction with Parametric and Non-Parametric Model Optimization
标题：FedCV ++：通过参数和非参数模型优化进行心血管风险预测的通信高效联邦学习
链接：https://arxiv.org/abs/2507.22963

【3】A Privacy-Preserving Federated Framework with Hybrid Quantum-Enhanced Learning for Financial Fraud Detection
标题：具有混合量子增强学习的隐私保护联邦框架用于金融欺诈检测
链接：https://arxiv.org/abs/2507.22908

推理|分析|理解|解释(3篇)

【1】Good Learners Think Their Thinking: Generative PRM Makes Large Reasoning Model More Efficient Math Learner
标题：好的学习者思考他们的想法：生成性PRM使大型推理模型更高效数学学习者
链接：https://arxiv.org/abs/2507.23317

【2】Multi-Hazard Early Warning Systems for Agriculture with Featural-Temporal Explanations
标题：具有特征-时间特征的农业多危害预警系统
链接：https://arxiv.org/abs/2507.22962

【3】Simulation-based inference for Precision Neutrino Physics through Neural Monte Carlo tuning
标题：通过神经蒙特卡罗调整对精确中微子物理进行基于模拟的推理
链接：https://arxiv.org/abs/2507.23297

检测相关(5篇)

【1】Manifold-regularised Signature Kernel Large-Margin $\ell_p$-SVDD for Multidimensional Time Series Anomaly Detection
链接：https://arxiv.org/abs/2507.23449

【2】Detection of Adulteration in Coconut Milk using Infrared Spectroscopy and Machine Learning
标题：使用红外光谱和机器学习检测椰奶中的掺假
链接：https://arxiv.org/abs/2507.23418

【3】Honey Adulteration Detection using Hyperspectral Imaging and Machine Learning
标题：使用高光谱成像和机器学习检测蜂蜜掺假
链接：https://arxiv.org/abs/2507.23416

【4】A Machine Learning Approach for Honey Adulteration Detection using Mineral Element Profiles
标题：使用矿物质元素谱检测蜂蜜掺假的机器学习方法
链接：https://arxiv.org/abs/2507.23412

【5】NaN-Propagation: A Novel Method for Sparsity Detection in Black-Box Computational Functions
标题：NaN传播：黑匣子计算函数稀疏性检测的一种新方法
链接：https://arxiv.org/abs/2507.23186

分类|识别(3篇)

【1】Impact of Hyperparameter Optimization on the Accuracy of Lightweight Deep Learning Models for Real-Time Image Classification
标题：超参数优化对实时图像分类轻量级深度学习模型准确性的影响
链接：https://arxiv.org/abs/2507.23315

【2】CNN-based solution for mango classification in agricultural environments
标题：基于CNN的农业环境芒果分类解决方案
链接：https://arxiv.org/abs/2507.23174

【3】Evaluating and Improving the Robustness of Speech Command Recognition Models to Noise and Distribution Shifts
标题：语音命令识别模型对噪声和分布漂移鲁棒性的评估和改进
链接：https://arxiv.org/abs/2507.23128

优化|敛散性(5篇)

【1】Differentially Private Clipped-SGD: High-Probability Convergence with Arbitrary Clipping Level
标题：差异私人剪辑-新元：任意剪辑水平的高概率收敛
链接：https://arxiv.org/abs/2507.23512

【2】Adjoint-Based Aerodynamic Shape Optimization with a Manifold Constraint Learned by Diffusion Models
标题：扩散模型学习的具有总管约束的基于伴随的气动形状优化
链接：https://arxiv.org/abs/2507.23443

【3】Investigating the Invertibility of Multimodal Latent Spaces: Limitations of Optimization-Based Methods
标题：研究多峰潜空间的可逆性：基于优化的方法的局限性
链接：https://arxiv.org/abs/2507.23010

【4】Optimal Transport Learning: Balancing Value Optimization and Fairness in Individualized Treatment Rules
标题：最佳交通学习：平衡个性化治疗规则中的价值优化和公平性
链接：https://arxiv.org/abs/2507.23349

【5】On the Complexity of Finding Stationary Points in Nonconvex Simple Bilevel Optimization
标题：非凸简单二层优化中寻找稳定点的复杂性
链接：https://arxiv.org/abs/2507.23155

预测|估计(4篇)

【1】FuseTen: A Generative Model for Daily 10 m Land Surface Temperature Estimation from Spatio-Temporal Satellite Observations
标题：FueTen：根据时空卫星观测估计每日10 m陆地表面温度的生成模型
链接：https://arxiv.org/abs/2507.23154

【2】A Foundation Model for Material Fracture Prediction
标题：材料断裂预测的基础模型
链接：https://arxiv.org/abs/2507.23077

【3】Prediction of Significant Creatinine Elevation in First ICU Stays with Vancomycin Use: A retrospective study through Catboost
标题：预测首次入住ICU时使用万科霉素后的肌群明显升高：通过Catboost进行的一项回顾性研究
链接：https://arxiv.org/abs/2507.23043

【4】Planning for Cooler Cities: A Multimodal AI Framework for Predicting and Mitigating Urban Heat Stress through Urban Landscape Transformation
标题：规划更凉爽的城市：通过城市景观改造预测和缓解城市热压力的多模式人工智能框架
链接：https://arxiv.org/abs/2507.23000

其他神经网络|深度学习|模型|建模(18篇)

【1】Consensus-Driven Active Model Selection
标题：问题驱动的主动模型选择
链接：https://arxiv.org/abs/2507.23771

【2】SHAP-Guided Regularization in Machine Learning Models
标题：机器学习模型中的SHAP引导正规化
链接：https://arxiv.org/abs/2507.23665

【3】On the Expressiveness of Softmax Attention: A Recurrent Neural Network Perspective
标题：论Softmax注意力的表现力：回归神经网络的视角
链接：https://arxiv.org/abs/2507.23632

【4】L-GTA: Latent Generative Modeling for Time Series Augmentation
标题：L-GTA：用于时间序列增强的潜在生成建模
链接：https://arxiv.org/abs/2507.23615

【5】EB-gMCR: Energy-Based Generative Modeling for Signal Unmixing and Multivariate Curve Resolution
标题：EB-gMCR：用于信号分解和多元曲线分辨率的基于能量的生成建模
链接：https://arxiv.org/abs/2507.23600

【6】Hardware-Aware Fine-Tuning of Spiking Q-Networks on the SpiNNaker2 Neuromorphic Platform
标题 ：SpiNNaker 2神经形态平台上尖峰Q网络的硬件感知微调
链接：https://arxiv.org/abs/2507.23562

【7】Continual Learning with Synthetic Boundary Experience Blending
标题：通过合成边界经验融合进行持续学习
链接：https://arxiv.org/abs/2507.23534

【8】Coflex: Enhancing HW-NAS with Sparse Gaussian Processes for Efficient and Scalable DNN Accelerator Design
标题：Expresslex：利用稀疏高斯过程增强HW-NAS，以实现高效且可扩展的DNN加速器设计
链接：https://arxiv.org/abs/2507.23437

【9】Scalable and Precise Patch Robustness Certification for Deep Learning Models with Top-k Predictions
标题：具有Top-k预测的深度学习模型的可扩展且精确的补丁鲁棒性认证
链接：https://arxiv.org/abs/2507.23335

【10】SequenceLayers: Sequence Processing and Streaming Neural Networks Made Easy
标题：SequenceLayers：简化序列处理和流神经网络
链接：https://arxiv.org/abs/2507.23292

【11】Evaluating the Dynamics of Membership Privacy in Deep Learning
标题：评估深度学习中会员隐私的动态
链接：https://arxiv.org/abs/2507.23291

【12】A Single Direction of Truth: An Observer Model's Linear Residual Probe Exposes and Steers Contextual Hallucinations
标题：真理的单向：观察者模型的线性剩余探测器揭露并引导情境幻觉
链接：https://arxiv.org/abs/2507.23221

【13】Model Directions, Not Words: Mechanistic Topic Models Using Sparse Autoencoders
标题：模型方向，而不是单词：使用稀疏自动编码器的机械主题模型
链接：https://arxiv.org/abs/2507.23220

【14】Learning to Prune Branches in Modern Tree-Fruit Orchards
标题：学习在现代树果园修剪树枝
链接：https://arxiv.org/abs/2507.23015

【15】Scientific Machine Learning with Kolmogorov-Arnold Networks
标题：基于Kolmogorov-Arnold网络的科学机器学习
链接：https://arxiv.org/abs/2507.22959

【16】Neural Autoregressive Modeling of Brain Aging
标题：大脑衰老的神经自回归建模
链接：https://arxiv.org/abs/2507.22954

【17】Scaled Beta Models and Feature Dilution for Dynamic Ticket Pricing
标题：动态票价的Scaled Beta模型和特征稀释
链接：https://arxiv.org/abs/2507.23767

【18】DNN-based Methods of Jointly Sensing Number and Directions of Targets via a Green Massive H2AD MIMO Receiver
标题：基于DNN的绿色大规模H2 AD MMO接收机联合感知目标数量和方向的方法
链接：https://arxiv.org/abs/2507.22906

其他(29篇)

【1】SUB: Benchmarking CBM Generalization via Synthetic Attribute Substitutions
标题：警告：通过合成属性替换对煤层气综合进行基准测试
链接：https://arxiv.org/abs/2507.23784

【2】DepMicroDiff: Diffusion-Based Dependency-Aware Multimodal Imputation for Microbiome Data
标题：DepMicroDiff：微生物组数据的基于扩散的依赖性感知多模式插补
链接：https://arxiv.org/abs/2507.23676

【3】One-Step Flow Policy Mirror Descent
标题：一步流程政策镜像下降
链接：https://arxiv.org/abs/2507.23675

【4】TweakLLM: A Routing Architecture for Dynamic Tailoring of Cached Responses
标题：TweakLLM：一种动态调整缓存响应的路由架构
链接：https://arxiv.org/abs/2507.23674

【5】Consistent Point Matching
标题：一致的点匹配
链接：https://arxiv.org/abs/2507.23609

【6】Optimised Feature Subset Selection via Simulated Annealing
标题：通过模拟安妮优化特征子集选择
链接：https://arxiv.org/abs/2507.23568

【7】Improved Algorithms for Kernel Matrix-Vector Multiplication Under Sparsity Assumptions
标题：稀疏性假设下的核矩阵-载体相乘改进算法
链接：https://arxiv.org/abs/2507.23539

【8】Transparent AI: The Case for Interpretability and Explainability
标题：透明的人工智能：可解释性和可解释性的案例
链接：https://arxiv.org/abs/2507.23535

【9】H-RDT: Human Manipulation Enhanced Bimanual Robotic Manipulation
标题：H-RDT：人类操纵增强的双手机器人操纵
链接：https://arxiv.org/abs/2507.23523

【10】A Verifier Hierarchy
标题：验证者层次结构
链接：https://arxiv.org/abs/2507.23504

【11】Directional Ensemble Aggregation for Actor-Critics
标题：演员-评论家的定向包围聚合
链接：https://arxiv.org/abs/2507.23501

【12】Merging Memory and Space: A Spatiotemporal State Space Neural Operator
标题：融合记忆和空间的时空状态空间神经算子
链接：https://arxiv.org/abs/2507.23428

【13】Causal Explanation of Concept Drift -- A Truly Actionable Approach
标题：概念漂移的因果解释--一种真正可行的方法
链接：https://arxiv.org/abs/2507.23389

【14】SWE-Exp: Experience-Driven Software Issue Resolution
标题：SWE-Exp：体验驱动的软件问题解决
链接：https://arxiv.org/abs/2507.23361

【15】SWE-Debate: Competitive Multi-Agent Debate for Software Issue Resolution
标题：SWE辩论：软件问题解决的竞争性多代理辩论
链接：https://arxiv.org/abs/2507.23348

【16】Designing Dynamic Pricing for Bike-sharing Systems via Differentiable Agent-based Simulation
标题：基于微分Agent的共享单车动态定价仿真设计
链接：https://arxiv.org/abs/2507.23344

【17】Efficient Machine Unlearning via Influence Approximation
标题：通过影响力逼近实现高效的机器去学习
链接：https://arxiv.org/abs/2507.23257

【18】Geak: Introducing Triton Kernel AI Agent & Evaluation Benchmarks
标题：Geak：引入Triton核心AI代理和评估基准
链接：https://arxiv.org/abs/2507.23194

【19】Observational Multiplicity
标题：观察多重性
链接：https://arxiv.org/abs/2507.23136

【20】RASL: Retrieval Augmented Schema Linking for Massive Database Text-to-SQL
标题：RASL：大规模数据库文本到SQL的检索增强模式链接
链接：https://arxiv.org/abs/2507.23104

【21】On the Sustainability of AI Inferences in the Edge
标题：论边缘人工智能推理的可持续性
链接：https://arxiv.org/abs/2507.23093

【22】Locally Differentially Private Thresholding Bandits
标题：局部差异性私有持票土匪
链接：https://arxiv.org/abs/2507.23073

【23】Linking Actor Behavior to Process Performance Over Time
标题：将演员行为与随着时间的推移的过程绩效联系起来
链接：https://arxiv.org/abs/2507.23037

【24】Data Readiness for Scientific AI at Scale
标题：大规模科学人工智能的数据准备
链接：https://arxiv.org/abs/2507.23018

【25】Stop Evaluating AI with Human Tests, Develop Principled, AI-specific Tests instead
标题：停止通过人体测试评估人工智能，转而开发有原则的、特定于人工智能的测试
链接：https://arxiv.org/abs/2507.23009

【26】CHECK-MAT: Checking Hand-Written Mathematical Answers for the Russian Unified State Exam
标题：CLARK-MAT：检查俄罗斯统一国家考试的手写数学答案
链接：https://arxiv.org/abs/2507.22958

【27】Extended Factorization Machine Annealing for Rapid Discovery of Transparent Conducting Materials
标题：扩展因子分解机退变快速发现透明导电材料
链接：https://arxiv.org/abs/2507.23160

【28】A Smoothing Newton Method for Rank-one Matrix Recovery
标题：一级矩阵恢复的光滑牛顿法
链接：https://arxiv.org/abs/2507.23017

【29】Typing Tensor Calculus in 2-Categories (I)
标题：输入2类张量微积分（I）
链接：https://arxiv.org/abs/1908.01212

机器翻译由腾讯交互翻译提供，仅供参考

点击“阅读原文”获取带摘要的学术速递