cs.AI（2026-10-01）

📊 共 75 篇论文 | 🔗 7 篇有代码

🎯 兴趣领域导航

支柱九：具身大模型 (Embodied Foundation Models) (43 🔗4) 支柱二：RL算法与架构 (RL & Architecture) (29 🔗3) 支柱三：空间感知与语义 (Perception & Semantics) (2) 支柱一：机器人控制 (Robot Control) (1)

🔬 支柱九：具身大模型 (Embodied Foundation Models) (43 篇)

#题目一句话要点标签🔗⭐
1 CoEvolve: Construct-to-Edit Visual Grounding with Bidirectional State Refinement 提出CoEvolve框架以解决视觉定位中的状态编辑问题 multimodal visual grounding ✅
2 Walking the Embedding Space: Datastore Extraction from Multimodal RAG 提出$ ext{immrag}$以解决多模态RAG中的数据提取攻击问题 large language model multimodal
3 Counterfactual Auditing of Bias in Open-Source Large Language Models for Clinical Triage 提出反事实审计方法以评估开源大语言模型在临床分诊中的偏见 large language model
4 External Observers May See More Clearly: Cross-Model Span-Level Hallucination Detection in Large Language Models via Hidden State Probing 提出跨模型的隐状态探测框架以解决大语言模型的幻觉检测问题 large language model
5 PRISM: A Category-Theoretic Framework for Measuring and Refining Multimodal Analogies 提出PRISM框架以解决多模态类比生成中的结构理解问题 multimodal
6 Gacha Decoding: Eliciting Diverse Generations Through Instruction Following 提出Gacha解码以解决生成多样性不足问题 instruction following
7 ReCast: Contract-Preserving Protection for Fixed-Interface Multimodal Reasoning 提出ReCast以解决多模态推理中的隐私保护问题 multimodal
8 CortexBridge: Cortical Alignment of EEG Montages for Foundation Models 提出CortexBridge以解决EEG电极布局变化带来的迁移困难 foundation model
9 AVSD-Scenes: A Dataset for Audio-Visual Description of Urban Scenes 提出AVSD-Scenes数据集以解决城市场景音视频描述不足问题 large language model multimodal
10 Token Communication-Assisted Collaborative Embodied Artificial Intelligence: Concepts, Framework, and Opportunities 提出Token通信框架以提升协作体智能的沟通效率 foundation model multimodal
11 VISTA: A Visual Harness for Reasoning in an Interactive World 提出VISTA以提升多模态模型在交互环境中的推理能力 multimodal
12 AiSearch: Interactive Multi-Modal Search with VLMs 提出AiSearch以解决多模态检索的交互性与自动化问题 multimodal
13 Causal Memory Policy: Making Memory Utility Identifiable by Intervening on Retrieval 提出因果记忆策略以解决记忆效用识别问题 large language model
14 HydroJEV: A one-second, training-free screen for cyber-attack and fault attribution in water distribution networks 提出HydroJEV以解决水分配网络中的网络攻击与故障归因问题 large language model
15 Mimir: Physics-Grounded LLM Agents for Long-Horizon Irrigation Control 提出Mimir以解决长时间段灌溉控制问题 large language model
16 Counting Moves, Weighing Voices: Bayesian Dialectical Argumentation for Calibrated Multi-LLM Councils under Persistent Adversaries 提出贝叶斯辩证论证以解决多LLM议会的可靠性问题 large language model
17 From Isolated Feature to Orbits: Discovering Music Concepts via Multi-SAE Alignment 提出基于多视角自编码器对齐的音乐概念发现方法 foundation model
18 Detecting Inconsistencies in Model Specifications with LLM-as-Verifier Reasoning 提出VeriSpec以解决模型规范不一致性问题 large language model ✅
19 TopK-Guided: Adaptive, Budget-Aware Activation Sparsity for Efficient LLM Inference 提出TopK-Guided以解决LLM推理中的激活稀疏性问题 large language model
20 FedLore: Communication and Memory Efficient Federated Learning via Shared Gradient Low-Rank Projection 提出FedLore以解决联邦学习中的通信与内存效率问题 foundation model
21 MCRI: A Four-Dimensional Framework for Analyzing and Evaluating Agent Skills 提出MCRI框架以系统分析和评估智能体技能 large language model
22 DRelay: Global Draft Context for Prefix-Aware Parallel Speculative Decoding Repair 提出DRelay以解决大语言模型的前缀选择错误问题 large language model
23 SpikeMoE: Brain-Inspired Competitive Routing for Flexible Spiking Mixture-of-Experts 提出SpikeMoE以解决灵活的脉冲混合专家选择问题 multimodal
24 PACE: Provenance-Aware Capability Enforcement for Tool-Using LLM Agents 提出PACE以解决工具使用LLM代理的安全性问题 large language model
25 Learning to Ask: Information Acquisition for SLM-LLM Collaboration, under a budget 提出信息获取方法以优化SLM-LLM协作 large language model
26 HHR: Hierarchical Hash Retrieval for Efficient LLM Generation 提出层次哈希检索以解决大语言模型生成效率问题 large language model ✅
27 Federated Agent Optimization 提出联邦智能体优化以解决隐私约束下的经验共享问题 large language model
28 Auditing Action Settlement in LLM Agent Environments: Order, Progress, and Replay 提出类型快照结算合约以解决LLM代理环境中的并发行动仲裁问题 large language model
29 Beyond State-of-the-Art: Standardising Environmental Impact Metrics for AI Research 提出标准化环境影响指标以推动可持续AI研究 large language model
30 Jev-IDS: System One Models for Network Intrusion Detection 提出JEV-IDS以解决网络入侵检测中的标签稀缺问题 large language model
31 MOMAT: Mixture of Multiple Atlases for Low-Power Jailbreak Defense of Quantized LLMs 提出MOMAT框架以解决量化大语言模型的监狱突破防御问题 large language model
32 Pay for the Fault, Not the Flow: Label-Free In-Flow Multi-Agent Workflow Optimization 提出InFlowOp以优化多智能体工作流，解决任务分解与评估问题 large language model ✅
33 TasteBench: Multimodal Benchmark for Sensory Prediction, from Molecules to Sustainable Foods 提出TasteBench以解决可持续蛋白质发现中的感官预测问题 multimodal
34 MemFit: Efficient Long-Term Agentic Memory 提出MemFit以解决长时记忆系统效率低下问题 large language model multimodal
35 Hypothesis-guided discovery of cognitive algorithms via program refinement 提出一种混合系统以解决认知算法发现问题 large language model
36 On-Premises Multi-Course RAG Tutoring for Business Education: Hardware-Software Trade-offs in a Campus AI Tutor 提出基于RAG的校园AI辅导系统以提升商业教育效果 large language model
37 What Does a Token Cost? A Mixture-of-Agents Measurement of Sufficient Per-Token Compute 提出混合代理模型测量每个令牌的计算成本以优化语言模型 large language model
38 Mitigating Private Data Leakage in LLMs with Whiteout 提出Whiteout以解决大型语言模型中的私人数据泄露问题 large language model
39 Choosing Before Acting: Comparative Value Estimation for Long-Horizon Tool-Use Agents 提出CITA以解决长时间工具使用中的价值估计问题 large language model
40 Causal Memory Policy: Making Memory Utility Identifiable by Intervening on Retrieval 提出因果记忆策略以解决记忆效用识别问题 large language model
41 Evaluating LLM-Generated Preference Distributions 系统分析LLM生成的偏好分布以提升决策支持 large language model
42 ABDA-NL: A Natural-Language Scenario Explorer for Argument-Based Reasoning 提出ABDA-NL以增强基于论证的自然语言推理能力 large language model
43 OR for AI That Does OR: Routing LLMs up the Escalator inside the OSCAR Framework 提出OSCAR框架以优化大语言模型的决策过程 large language model

🔬 支柱二：RL算法与架构 (RL & Architecture) (29 篇)

#题目一句话要点标签🔗⭐
44 Decision Titan: Test-Time Training for Long-Term Memory in Offline Reinforcement Learning 提出Decision Titan以解决离线强化学习中的长期记忆问题 reinforcement learning offline RL offline reinforcement learning
45 Homomorphic Advantage Operator: Stabilizing Reinforcement Learning Under Fully Homomorphic Encryption Constraints 提出同态优势算子以解决强化学习中的同态加密问题 reinforcement learning OMOMO
46 SPHERE: Adaptive VR Indoor Scene Generation via LLM-Enhanced Spatial Preference Learning and Human-in-the-Loop RL 提出SPHERE以解决用户偏好在VR室内场景生成中的不足问题 reinforcement learning preference learning large language model ✅
47 Dependency-Aware Reward Shaping for Agentic Reinforcement Learning 提出依赖感知奖励塑形方法以优化强化学习中的步骤信用分配 reinforcement learning reward shaping large language model ✅
48 On the Divergence of Accuracy and Mechanism Consistency in Time Series World Models 提出时间序列世界模型以解决预测准确性与机制一致性问题 world model world models MAE
49 Grounding Large Language Models in DSGE Simulators for Policy Generation and Forecasting 将大型语言模型与DSGE模拟器结合以生成经济政策 PPO large language model
50 Network World Models as Environments for Algorithm Design on Complex Systems 提出基于网络世界模型的算法设计方法以优化复杂系统干预 world model world models
51 Optimal Transport Meets Reinforcement Learning: A Survey 综述最优传输在强化学习中的应用与挑战 reinforcement learning offline RL imitation learning
52 LineupRL: Verifiable Reinforcement Learning for Time Series Captioning via Caption-to-Series Identification 提出LineupRL以解决时间序列描述中的奖励设计问题 reinforcement learning large language model
53 PPO-HRAP: Proximal Policy Optimization with a Hybrid Regime-Aware Policy for Risk-Controlled Trading 提出PPO-HRAP以解决风险控制交易中的收益与回撤平衡问题 reinforcement learning PPO
54 Reinforcement Learning to Accelerate Primal-Dual Hybrid Gradient for Linear Programming 提出GALLOP以加速线性规划的原始-对偶混合梯度方法 reinforcement learning
55 Sharpening Tax in Post-Training 提出Sharpening Tax以量化后训练对解决方案覆盖率的影响 reinforcement learning large language model
56 Mapping the RAG Landscape: A Four Axis Taxonomy of Efficiency, Defense, Interactivity, and Reasoning 提出四轴分类法以提升检索增强生成模型的效率与可靠性 reinforcement learning large language model
57 Rethinking Probability-Based Reinforcement Learning From Posterior Concentration 提出RLCPR以解决长时推理中的奖励不稳定问题 reinforcement learning
58 ProtoFlow: Prototype-Guided Flow Matching for Multivariate Time Series Forecasting 提出ProtoFlow以解决多变量时间序列预测中的生成建模问题 flow matching
59 AbsorbEvo: An Agentic Framework for Autonomous Inverse Design of Microwave Absorbers 提出AbsorbEvo框架以实现微波吸收体的自主逆向设计 predictive model large language model
60 Improving Math Reasoning through Value-guided Informative Search 提出APIVIS框架以提升数学推理能力 reinforcement learning large language model
61 ActiveSaddler: Automated Curriculum Learning for Agent Harness Optimization 提出ActiveSaddler以解决自动化课程学习中的优化问题 curriculum learning
62 IGNITE Tokamak World Model Architecture 提出IGNITE模型以高效模拟核聚变等离子体行为 world model world models foundation model
63 Calibration-risk routing for controlled world-model adaptation 提出模型校正世界模型以解决模拟器与目标之间的适应问题 reinforcement learning world model world models
64 World Action Modeling with Progressive Visual Planning 提出ProWAM以解决长时间预测中的视觉动态与动作生成问题 world action model world action models ✅
65 How To Train Your World Model: Fine-tuning vs RAG for LM-based World Modeling 提出混合世界建模系统以优化语言模型的环境模拟 world model world models
66 Tropical Reinforcement Learning 提出热带强化学习以解决语言模型的组合推理问题 reinforcement learning large language model
67 RISED: RubrIcs for agentic multi-environment Selection and sElf-Distillation 提出RISED以解决多环境强化学习中的数据选择问题 distillation generalist agent
68 THPL: A Vision-to-Language Decision Support Framework for Rainbow Trout Feeding Management in RAS 提出THPL框架以解决循环水养殖系统中虹鳟鱼喂养管理问题 direct preference optimization spatiotemporal multimodal
69 Reinforcement Learning Techniques for the Optimization of Target Polarization in Nuclear Physics Scattering Experiments 提出数据驱动控制框架以优化核物理散射实验中的目标极化 reinforcement learning
70 Rethinking Probability-Based Reinforcement Learning From Posterior Concentration 提出RLCPR以解决长序列推理中的奖励不稳定问题 reinforcement learning
71 Learning Multiple Timescales for Goal-Conditioned Reinforcement Learning 提出GITA以解决长时间跨度目标条件强化学习问题 reinforcement learning
72 World Editing: Intervening on Executable Worlds at Increasing Depth 提出世界编辑方法以解决可执行世界编辑不足问题 world model world models

🔬 支柱三：空间感知与语义 (Perception & Semantics) (2 篇)

#题目一句话要点标签🔗⭐
73 NextMe-800: Anticipating Personal Behavior from Months of Egocentric Video 提出NextMe-800以解决个体行为预测问题 open-vocabulary open vocabulary egocentric
74 Automating the Application of HCI Principles: Skills for On-Demand UI Construction, the Human-AI Space to Think, and the Future of HCI 提出基于AI的HCI原则自动应用框架以提升用户界面构建效率 affordance large language model

🔬 支柱一：机器人控制 (Robot Control) (1 篇)

#题目一句话要点标签🔗⭐
75 Measuring the Stability Assumption Behind Action Chunking 研究动作分块背后的稳定性假设以提升行为克隆性能 manipulation imitation learning representation learning

⬅️ 返回 cs.AI 首页 · 🏠 返回主页