cs.LG（2026-10-13）

📊 共 41 篇论文 | 🔗 9 篇有代码

🎯 兴趣领域导航

支柱二：RL算法与架构 (RL & Architecture) (26 🔗7) 支柱九：具身大模型 (Embodied Foundation Models) (11 🔗1) 支柱一：机器人控制 (Robot Control) (3 🔗1) 支柱五：交互与反应 (Interaction & Reaction) (1)

🔬 支柱二：RL算法与架构 (RL & Architecture) (26 篇)

#题目一句话要点标签🔗⭐
1 SYNPRED: A Synergistic Approach to Multimodal Learning for Clinical Prediction 提出SYNPRED以解决多模态临床预测中的信息利用不足问题 contrastive learning multimodal ✅
2 Failure Recovery via Belief Updates and World Modeling for Long-Horizon Agents 提出BELIEFWEAVER框架以解决长时间跨度代理的故障恢复问题 world model world models distillation
3 CoCA: Criterion-Aware Credit Assignment for Rubric-Based Reinforcement Learning 提出CoCA以解决标准化反馈在强化学习中的不足问题 reinforcement learning distillation privileged information
4 DisaggKD: Efficient Knowledge Distillation through Teacher-Student Disaggregation and a Fused Loss 提出DisaggKD以解决知识蒸馏中的内存限制问题 teacher-student distillation
5 RELAY: Selective Self-Distillation with Adaptive Hand-off for Rubric-Based Reinforcement Learning 提出RELAY以解决基于评分标准的强化学习中的奖励梯度问题 reinforcement learning distillation
6 SkillScaffold: Scaffolding Agentic Reinforcement Learning with Policy-Adaptive Skill Credit Assignment and Selective Injection 提出SkillScaffold以解决技能注入不稳定性问题 reinforcement learning large language model ✅
7 ProAS: A Progress-Aware Advantage Shaping Framework for Multi-Turn Agentic Reinforcement Learning 提出ProAS框架以解决多回合强化学习中的信用分配问题 reinforcement learning large language model ✅
8 Does self-distillation improve reasoning? 重新审视自蒸馏在推理中的有效性 reinforcement learning distillation privileged information ✅
9 Stein Variational Distributional Reinforcement Learning: Score-Guided Particles for Multivariate Return Distributions 提出Stein变分分布强化学习以解决多元回报分布问题 reinforcement learning multimodal
10 TSRL: Trigger-based Supervised Reinforcement Learning for Continual Agent Training 提出TSRL框架以解决持续学习中的灾难性遗忘问题 reinforcement learning large language model
11 Learning When to Transfer: Reinforcement Learning for Carbon-Footprint Modeling in New Manufacturing Domains 提出强化学习方法以优化新制造领域的碳足迹建模 reinforcement learning predictive model
12 Teach Me After You Learn It: Lookahead Self-Distillation for RLVR 提出Lookahead自蒸馏方法以提升RLVR性能 distillation privileged information
13 When Does Opponent Modeling Pay Off? Spatial Inference Stabilises Reinforcement Learning under Partial Observability 提出空间对手建模以稳定部分可观测强化学习 reinforcement learning deep reinforcement learning
14 Explore with Memory, Learn without It: In-Context Exploration and Self-Distillation for LLM Reasoning 提出LEAF以解决强化学习中的探索效率问题 reinforcement learning distillation ✅
15 Two Clocks: Why Larger Rollout Batches Learn Slower but Fail on Schedule in Reinforcement Learning with Verifiable Rewards 提出两时钟框架以解决强化学习中样本效率问题 reinforcement learning ✅
16 Deterministic Prediction as Global Velocity: Flow Matching for Probabilistic Time-Series Forecast 提出DV-Flow以解决时间序列预测中的不确定性量化问题 flow matching ✅
17 Euclidean Distance-Based Flow Matching 提出基于欧几里得距离的流匹配以解决数据污染问题 flow matching
18 Teacher-Guided Tool Learning for Small Language Models via Bidirectional KL Constraints 提出选择性教师指导策略优化以解决小语言模型工具学习问题 reinforcement learning teacher-student distillation
19 Reinforcement Learning for Large-Scale Lifelong Multi-Agent Path Finding with Shared Spatial Fields 提出一种基于强化学习的多智能体路径规划方法以应对复杂交通环境 reinforcement learning
20 LoOPSD: On-Policy Self-Distillation with Looped Teachers 提出LoOPSD以通过层执行顺序优化自我蒸馏 distillation
21 FRACTION: Fraud Explanation Generation via Agentic Graph Context Navigation 提出FRACTION框架以解决图结构欺诈解释生成问题 distillation large language model
22 Correcting or Continuing? What On-Policy Distillation Learns from Student Errors 探讨在政策蒸馏中如何应对学生错误 distillation
23 Improving Neural Symbolic Regression with a Diverse Prior and Post-training Alignment 提出RASR以提升神经符号回归在真实数据上的表现 reinforcement learning foundation model
24 Checking the Checkers: A Synthesis of Verifier Noise and Robustness in RLVR 提出对RLVR中验证器噪声与鲁棒性的新综合分析 reinforcement learning large language model
25 Posterior-Averaged Bellman Flow Matching for Distributional Critics 提出后验平均贝尔曼流匹配以优化分布式评论者 flow matching
26 Enabling Token-wise Importance Sampling in Off-Policy RLVR with Proximal KL Regularization 提出Proximal KL方法以解决离线强化学习中的重要性采样问题 reinforcement learning PPO

🔬 支柱九：具身大模型 (Embodied Foundation Models) (11 篇)

#题目一句话要点标签🔗⭐
27 MMTA: Benchmarking Multimodal Temporal Analysis with Time Series, Text, and Vision 提出MMTA基准以解决多模态时间序列分析问题 large language model multimodal
28 HaloBlock: Faster and Better Decoding for Diffusion Large Language Models 提出HaloBlock以解决扩散大语言模型解码难题 large language model ✅
29 SP$^2$ec: Adaptive Self-Speculative Decoding for Vision-Language Models 提出SP$^2$ec以解决视觉语言模型推理效率问题 large language model
30 MultiAlloc: Budget-Aware Precision Allocation for KV Cache Quantization 提出MultiAlloc以解决KV缓存量化中的预算分配问题 large language model
31 How Low Can You Go: Fine-Tuning LLMs with Bayesian Optimization (LlMBO) 提出LlMBO以优化低参数LLM微调问题 large language model
32 Kairos: Transactional Security Enforcement at the System Boundary for LLM Agents 提出Kairos以解决LLM代理的事务安全问题 large language model
33 PRISM: Calibrating Black-Box Judges with Counterfactual Probes 提出PRISM以校准黑箱评估者的评分偏差问题 large language model
34 A Few More Tokens Ahead: Reinforcing Parallelism in Diffusion LLMs 提出RePara以增强扩散大语言模型的并行生成能力 large language model
35 BELT : BEhavior-Aligned Low-Rank KV Cache Compression with Learned Top-k Rank Allocation 提出BELT框架以解决长上下文推理中的KV缓存瓶颈问题 large language model
36 FlashMask-3: Efficient and Expressive Mask-Aware Distributed Attention 提出FlashMask-3以解决长上下文训练中的注意力掩码瓶颈问题 multimodal
37 Agentic-MatriBO: Capability-Compiled Agentic Bayesian Optimization for Materials and Chemical Design 提出Agentic-MatriBO以解决材料与化学设计中的优化问题 large language model

🔬 支柱一：机器人控制 (Robot Control) (3 篇)

#题目一句话要点标签🔗⭐
38 Where Small Mistakes Matter: Quantifying State Criticality for Imitation Learning 提出SCRIM以量化模仿学习中的状态关键性问题 manipulation imitation learning behavior cloning ✅
39 Model-Based Online Decision Making via Generative Trajectory Planning 提出生成轨迹规划以解决在线决策中的长远推理问题 manipulation trajectory optimization diffusion policy
40 Better Capturing Interactions Between Products in~Retail: Revisited Negative Sampling for~Basket Choice Modeling 提出基于负采样的购物篮模型以优化零售产品组合 motion planning

🔬 支柱五：交互与反应 (Interaction & Reaction) (1 篇)

#题目一句话要点标签🔗⭐
41 Privacy-Preserving GNN Inference on Financial Transaction Graphs via Homomorphic Encryption 提出隐私保护的GNN推理框架以解决金融交易图的隐私问题 OMOMO

⬅️ 返回 cs.LG 首页 · 🏠 返回主页