cs.LG（2026-10-05）

📊 共 56 篇论文 | 🔗 5 篇有代码

🎯 兴趣领域导航

支柱二：RL算法与架构 (RL & Architecture) (25 🔗3) 支柱九：具身大模型 (Embodied Foundation Models) (24 🔗2) 支柱一：机器人控制 (Robot Control) (5) 支柱四：生成式动作 (Generative Motion) (2)

🔬 支柱二：RL算法与架构 (RL & Architecture) (25 篇)

#题目一句话要点标签🔗⭐
1 EpicWorldModel: Exploration-driven Planning with Latent World Models 提出EpicWorldModel以解决部分可观测环境中的不确定性规划问题 flow matching world model worldmodel
2 Ramp Metering Control via Hybrid State Deep Reinforcement Learning in Partially Observable Connected Vehicle Environments 提出混合状态深度强化学习以解决部分可观测环境下的匝道控制问题 reinforcement learning deep reinforcement learning DRL ✅
3 Conditional Flow Matching for Single-Neuron Electrophysiology: Capturing Multimodal Responses Across Stimuli 提出条件流匹配模型以解决单神经元电生理响应多样性问题 flow matching multimodal
4 Ultrasound Operator Guidance Using World Modeling and Retrieval Based Action Planning 提出基于世界建模和检索的超声操作指导方法以提升成像质量 world model world models JEPA
5 Beyond In-Distribution Preservation: Recovering Generalization in Quantized VLAs via Vulnerability-Oriented Tuning 提出PIVOT-Q以解决量化VLA的鲁棒性与泛化问题 distillation VLA ✅
6 What Matters for Latent Reasoning with Flow Matching 提出FLaRe以解决潜在推理效率与准确性问题 flow matching large language model chain-of-thought
7 Considering Context: When World Models Need Context Encoders 提出预测充分性以优化模型驱动强化学习中的上下文编码 reinforcement learning world model world models
8 Generative World Models Enable Predictive Control of Laser Melt Pool Dynamics 提出生成世界模型以实现激光熔池动态的预测控制 world model world models latent dynamics
9 Boosting Transferable Adversarial Attacks against Deep Reinforcement Learning 提出基于轨迹的攻击方法以增强深度强化学习的可转移对抗攻击 reinforcement learning deep reinforcement learning DRL
10 From Pixels, Without Pre-training: Joint Generative and Self-Supervised Representation Learning in One Model 提出SCION以解决无标签生成与自监督表示学习问题 flow matching representation learning contrastive learning
11 ARO: Aligned Representation learning for multi-Omics data 提出ARO以解决多组学数据缺失和不匹配问题 representation learning foundation model
12 Adaptive Expert Guidance for Efficient On-Policy Reinforcement Learning 提出自适应专家指导以提高在线强化学习的效率 reinforcement learning PPO
13 Transfer-Stratified On-Policy Distillation for RL-Improved Reasoning Teachers 提出转移分层的在线蒸馏方法以提升推理教师的效果 reinforcement learning distillation
14 ThunderSyncRL: Lossless Acceleration of Agentic Reinforcement Learning 提出ThunderSyncRL以解决强化学习中的同步效率问题 reinforcement learning distillation
15 Beyond Transport Cost: Routing Differences between Flow Matching and Optimal Transport 提出基于路由的流匹配方法以优化生成模型 flow matching
16 Sharpen Without Search: On-Policy Distillation of Sequence-Level Power Distribution 提出在政策蒸馏中优化序列级功率分布的方法 distillation ✅
17 Hyperbolic Graph Representation Learning: Embed in One Metric, Optimize with Another 提出超曲率图表示学习以优化嵌入与学习过程 representation learning
18 WaveGSSM: Graph Wave State Space Models for Propagating Spatio-Temporal Patterns 提出WaveGSSM以解决时空图模式传播问题 state space model
19 FairProp: Fair Node Representation Learning via Differentiable Propagation Layers 提出FairProp以解决图神经网络中的公平性问题 representation learning
20 Constrained Goal-directed Planar Graph Generation with Grammar-based Reinforcement Learning 提出基于语法的强化学习方法以解决平面图生成问题 reinforcement learning
21 Reinforcement Learning-Based Optimization of Workload-Aware Power Delivery Networks 提出基于强化学习的优化方法以解决PDN设计效率问题 reinforcement learning
22 Flash-OPD: Fast On-Policy Distillation 提出Flash-OPD以解决长轨迹生成的高训练成本问题 distillation
23 MercerFlow: Flow Matching in a Kernel-Induced Latent Space for Probabilistic Forecasting 提出MercerFlow以解决时间序列预测中的流匹配问题 flow matching
24 Off-Policy Merging Beats On-Policy Self-Distillation for Continual Learning 提出离线合并方法以超越在线自蒸馏解决持续学习问题 distillation
25 Decoupling Time and Space: A Temporally Conditioned Refinement for EEG Source Imaging 提出双流框架以解决EEG源成像中的时空耦合问题 representation learning spatiotemporal

🔬 支柱九：具身大模型 (Embodied Foundation Models) (24 篇)

#题目一句话要点标签🔗⭐
26 OGAM: Connecting Systematic Testing to Runtime Assurance through Object-Grounded Attention Monitoring for VLA Policies 提出OGAM以解决VLA策略的运行保障问题 vision-language-action VLA OpenVLA
27 Aligning Multimodal Patient Evidence with Biomedical Knowledge Graphs for Clinical LLMs 提出MM-KG以解决临床问题与生物医学知识图谱的对接问题 large language model multimodal
28 MatrixFormer: A Foundation Model for Matrix Completion 提出MatrixFormer以解决矩阵补全问题 foundation model
29 Xaurora: Generative Weather Forecasting with Denoising Stochastic Interpolants from a Foundation Model Prior 提出Xaurora以解决天气预报中的不确定性量化问题 foundation model
30 EMG-FM-Bench: A Comprehensive Benchmark for Foundation Model Transfer and Adaptation on Electromyography 提出EMG-FM-Bench以解决肌电图模型迁移与适应性问题 foundation model
31 Scaling Down the Scaling Laws: Parameter Efficiency and Compute-Optimal Training in Resource-Constrained Large Language Models 提出资源高效的训练框架以应对大语言模型的扩展挑战 large language model
32 Multimodal Deep Survival Analysis for Sinkhole Susceptibility 提出多模态Cox比例风险模型以解决塌陷易发性预测问题 multimodal
33 SimAuthor: Harnessing Foundation Models for Persistent Scientific Simulator Authoring 提出SimAuthor以解决科学模拟器的持续作者问题 foundation model
34 Anlu: Enabling In-Context Time Series Anomaly Detection in Foundation Models via Counterfactual Supervision 提出Anlu以解决时间序列异常检测中的上下文学习问题 foundation model
35 On the Geometry of Multimodal Saturation: Riemannian VICReg 提出Riemannian VICReg以解决多模态饱和问题 multimodal
36 Correct Verdicts, Flawed Reasoning: Structured Auditing of LLM-based Vulnerability Reasoning 提出VERA框架以解决LLM漏洞推理的审计问题 large language model chain-of-thought
37 CCQ: A Multi-State Child Care Quality Dataset to Support AI for Children's Health Research 提出CCQ数据集以解决儿童保育质量研究数据不足问题 foundation model zero-shot transfer
38 How to scale your HEP ML models: A recipe for robust architecture comparisons at scale 提出系统化方法以优化高能物理机器学习模型的规模比较 foundation model
39 To Learn is to Wander: Learning Across Graphs and Tasks with Random Walks 提出Wander模型以解决图学习跨任务与跨图的挑战 foundation model
40 OVAL: Output-Aware Local Page Bases for KV Cache Retrieval 提出OVAL以解决KV缓存检索中的近似误差问题 large language model ✅
41 Closing the Context Gap: Activation Alignment for Tabular In-Context Learning 提出激活对齐方法以解决表格模型的上下文成本问题 foundation model
42 Separators Make Carry Propagation Learnable:The Geometry of Latent Carry in a Multiplication Transformer 提出分隔符使得乘法变换器中的进位传播可学习 chain-of-thought
43 Quantifying the Stability of Multi-Step Reasoning via Error Amplification 量化多步推理稳定性以解决错误放大问题 chain-of-thought
44 RoSA: Rotational Sparse Adaptation for Memory-Efficient Fine-Tuning 提出RoSA以解决内存效率低下的微调问题 foundation model
45 Certification-Enhanced Generalization Bounds 提出认证增强的泛化界限以提升学习算法的可靠性 large language model
46 TIGER: Time-Series Classification with In-Context-Learning Gated Ensemble of Representations 提出TIGER以解决时间序列分类中的适应性不足问题 foundation model
47 StagQ: Constraint-Driven Multi-Precision Weight Quantization for LLMs 提出StagQ以解决大语言模型的多精度权重量化问题 large language model
48 Physics-Informed but Not Physics-Consistent: Error Geometry and Subspace Projection for Neural AC Power Flow 提出校准子空间投影以解决神经AC潮流的物理一致性问题 foundation model ✅
49 CIPHER-MoE: Balancing Efficiency and Routing Fidelity in Trillion-Scale MoE Training 提出CIPHER-MoE以解决大规模MoE训练中的负载不均问题 large language model

🔬 支柱一：机器人控制 (Robot Control) (5 篇)

#题目一句话要点标签🔗⭐
50 H-JEPA: End-to-End Learning of Hierarchical World Models for Visual Planning 提出H-JEPA以解决长时间规划中的层次世界模型问题 manipulation world model world models
51 Time-series Foundation Models for Predictive Control: The Role of Excitation 提出时间序列基础模型以优化预测控制中的激励问题 MPC model predictive control predictive model
52 Reachability-Aware Diffusion Policy Optimization 提出可达性感知扩散策略优化以解决安全性问题 reachability-aware reinforcement learning diffusion policy
53 Structured Representation Learning for Behavior Cloning: How can we learn to safely control a nuclear power plant? 提出结构化表示学习以安全控制核电站 model predictive control behavior cloning representation learning
54 The Optimization Landscape of Learning Compacted Context Models 提出简化的Perceiver架构以优化持续学习中的KV缓存压缩问题 manipulation

🔬 支柱四：生成式动作 (Generative Motion) (2 篇)

#题目一句话要点标签🔗⭐
55 CACFG: Curvature-Aware Classifier-Free Guidance and Optimal Control 提出CACFG以解决CFG在高引导强度下的质量下降问题 classifier-free guidance
56 PhaseMatcher: Autoregressive Phase-Set Identification with Spectral Decomposition 提出PhaseMatcher以解决PXRD中相位识别问题 physically plausible

⬅️ 返回 cs.LG 首页 · 🏠 返回主页