cs.LG（2026-09-24）

📊 共 28 篇论文 | 🔗 1 篇有代码

🎯 兴趣领域导航

支柱九：具身大模型 (Embodied Foundation Models) (15 🔗1) 支柱二：RL算法与架构 (RL & Architecture) (11) 支柱一：机器人控制 (Robot Control) (1) 支柱五：交互与反应 (Interaction & Reaction) (1)

🔬 支柱九：具身大模型 (Embodied Foundation Models) (15 篇)

#题目一句话要点标签🔗⭐
1 ICE: Task-Aligned Clifford Latent Fields for Multimodal Graph Foundation Models 提出ICE模型以解决多模态图基础模型的表示问题 foundation model multimodal
2 TraceGuard: Adaptive Multimodal Poison Filtering through Cross-Feature Rank Agreement 提出TraceGuard以解决多模态数据中毒问题 multimodal
3 Automatic Rank Allocation for Low-Rank Adaptation in Large Language Models via lp Regularization 提出$ ext{l}_p$-LoRA以解决低秩适应中的秩分配问题 large language model
4 Beyond Average Safety: Chance-Constrained LLM Fine-tuning 提出机会约束方法以提升大语言模型的安全性 large language model instruction following
5 SwitchPFN: Shared Switching Dynamics for Frozen In-Context Time Series Classification 提出SwitchPFN以解决时间序列分类中的表示设计问题 foundation model
6 Edge AI on Constrained Devices for Binary Sleep-Wake Classification in Dynamic Environments 提出基于边缘AI的系统以解决动态环境中的睡眠状态分类问题 multimodal
7 Downside-Controlled Online Forecast Combination under Delayed and Revised Outcomes 提出下行控制的在线预测组合方法以应对延迟和修正结果问题 foundation model
8 EvoTreeNAD: Genealogy-Guided Evolution for LLM-Driven Neural Architecture Discovery 提出EvoTreeNAD以解决神经架构发现中的探索与评估问题 large language model
9 Reliability-aware Cross-sample Enhancement for Robust Multimodal Sentiment Analysis 提出可靠性感知的跨样本增强框架以解决多模态情感分析中的噪声问题 multimodal
10 In-Context Binding Capacity in Language Models 提出语言模型的上下文绑定能力测量方法 instruction following
11 Entropy Regularization: A Free Correction to Cross-Entropy for Verified Demonstrations 提出熵正则化以解决交叉熵在可验证任务中的不足 large language model
12 To Solve Bilevel Optimization with Nonconvex Lower Levels, We Need Second-Order Stationarity 提出PROBE算法以解决非凸下层双层优化问题 large language model
13 Adaptive Multi-Value Control in LLMs via Causal Activation Steering 提出AIMES框架以实现大语言模型的自适应多价值控制 large language model
14 Cost-Aware Best-LLM Identification using Dueling Feedback 提出基于对抗反馈的成本感知最佳LLM识别方法 large language model
15 Parameters vs. Context: TRACE Fine-Tuning for Robust Retrieval-Augmented Generation 提出TRACE以解决知识冲突下的RAG模型鲁棒性问题 large language model ✅

🔬 支柱二：RL算法与架构 (RL & Architecture) (11 篇)

#题目一句话要点标签🔗⭐
16 Aim Short to Reach Far: Your Frozen World Model Can Plan Better Than You Think 提出锚定规划以优化视觉世界模型的控制策略 world model world models
17 Graph-Based Inference and Topology-Aware Multi-Agent Reinforcement Learning for Large-Scale Railway Network Management 提出图基推理与拓扑感知多智能体强化学习以解决铁路网络管理问题 reinforcement learning deep reinforcement learning zero-shot transfer
18 PoEM: Predicting RL Outcomes from Existing Policies 提出PoEM框架以预测强化学习结果，降低计算成本 reinforcement learning foundation model instruction following
19 Temporal Gradient Inversion for Private Trajectory Reconstruction in Embodied Reinforcement Learning 提出TRACE以解决隐私保护下的轨迹重建问题 reinforcement learning policy learning
20 Learning and interpreting policies for simultaneous entanglement requests in quantum networks 提出基于强化学习的策略以解决量子网络中的纠缠请求调度问题 reinforcement learning large language model
21 Physics and Data Driven Transformer-Mamba Framework for Flow Field 提出Transformer-Mamba框架以解决流场计算中的物理一致性问题 Mamba
22 Not Every Token Is Worth Distilling: Selective Supervision for Direct-OPD 提出选择性监督方法S$^2$D-OPD以提升直接策略蒸馏效果 reinforcement learning distillation
23 Probabilistic Robustness-driven Universal Adversarial Perturbations with Explainability against Deep Reinforcement Learning-based Intrusion Detection System 提出基于概率鲁棒性的普适对抗扰动以增强深度强化学习入侵检测系统的安全性 reinforcement learning deep reinforcement learning DRL
24 Aim Short to Reach Far: Your Frozen World Model Can Plan Better Than You Think 提出锚定规划以提升视觉世界模型的控制能力 world model world models
25 From Weak Data to Strong Policy: Q-Targets Enable Provable In-Context Reinforcement Learning 提出Q-目标预训练变换器以解决弱数据下的强化学习问题 reinforcement learning behavior cloning
26 AutoResearch at Production Scale: Failure Modes and a Multi-Agent Framework 提出AutoResearch框架以解决生产推荐系统中的失败模式问题 representation learning large language model

🔬 支柱一：机器人控制 (Robot Control) (1 篇)

#题目一句话要点标签🔗⭐
27 Learning from Mixed-Quality Deployment Experience for Robot Manipulation 提出PACL以解决机器人操作中的混合质量经验学习问题 manipulation reinforcement learning offline reinforcement learning

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

#题目一句话要点标签🔗⭐
28 Encryptability As a Coordinate Choice: Depth-One Homomorphic Federated Learning of Quantum Neural Networks 提出一种新方法以实现量子神经网络的同态加密联邦学习 OMOMO

⬅️ 返回 cs.LG 首页 · 🏠 返回主页