cs.AI（2026-08-04）

📊 共 62 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) (46 🔗3) 支柱二：RL算法与架构 (RL & Architecture) (14 🔗4) 支柱一：机器人控制 (Robot Control) (2)

🔬 支柱九：具身大模型 (Embodied Foundation Models) (46 篇)

#题目一句话要点标签🔗⭐
1 Failure-Informed Image Self-Augmentation for Multimodal Large Language Model Self-Improvement 提出FISA框架以解决多模态大语言模型自我改进问题 large language model multimodal
2 Pattern over Pixels: Measuring Pattern Completion Bias in Multimodal Code Generation 提出视觉模式完成偏差基准以提升多模态代码生成准确性 large language model multimodal
3 KnowHal: A Knowledge-Driven Benchmark for Comprehensive Multimodal Hallucination Evaluation 提出KnowHal以解决多模态模型的知识幻觉评估问题 large language model multimodal
4 Taming the Implicit: Dual-Channel Risk-Aware Reinforcement Fine-Tuning for Continual Multimodal Post-Training 提出风险感知策略优化以解决多模态强化微调中的遗忘问题 large language model multimodal
5 Balancing Efficiency and Efficacy: Training-Free Attention-Guided Switching Between Explicit and Latent Thoughts for MLLMs 提出训练无关的注意力引导切换以解决多模态大语言模型推理问题 large language model multimodal chain-of-thought ✅
6 AI World Cup 2026: Benchmarking Large Language Models for End-to-End Football Tournament Prediction 提出AI世界杯基准以评估大型语言模型的足球赛事预测能力 large language model
7 Minimax-Optimal Semiparametric Contextual Dynamic Pricing with Multimodal Revenue 提出一种最小极大半参数上下文动态定价方法以解决多模态收益问题 multimodal
8 Can Large Language Models Recover Semantic Optimization Opportunities That Compilers Miss? 提出SeGaBench以利用大语言模型优化编译器遗漏的语义 large language model
9 A game theory for foundation models shows new paths to rational cooperation through similarity inference 提出嵌入式贝叶斯代理模型以解决AI代理合作问题 foundation model
10 A Security-Oriented Lifecycle Model for Large Language Model Systems 提出安全导向的生命周期模型以解决大语言模型系统的安全问题 large language model
11 Large language models for partial differential equation workflows 利用大型语言模型优化偏微分方程工作流 large language model
12 Reversing Arrows in Large Language Models 系统研究大型语言模型中的逆关系方向性问题 large language model
13 Adaptive Modality Reliability Diagnosis and Restoration for Robust Multimodal Intent Recognition 提出PRIME框架以解决多模态意图识别中的可靠性问题 multimodal
14 Screenshots or Tools? Eliciting Tool Use and Managing Multimodal Context in Hybrid GUI-MCP Computer-Use Agents 提出混合计算机使用代理以优化工具使用与多模态上下文管理 multimodal
15 Evidence-Grounded Multimodal Knowledge Graph Construction for Multi-Lecture Educational Reasoning 提出基于证据的多模态知识图谱构建方法以解决教育推理问题 multimodal
16 Separating quantum circuits from classical LLMs 提出量子电路与经典大语言模型的分离方法 large language model chain-of-thought
17 Does Forgetting Transfer Across Modalities? A Real-World Benchmark for Cross-Modal Knowledge Unlearning Evaluation 提出UNLINK-VL基准以解决跨模态知识遗忘评估问题 large language model multimodal
18 Soft Guidance Starts to Outperform CoT Prompting as LLMs Improve 提出软引导方法以超越传统CoT提示在推理任务中的表现 large language model chain-of-thought
19 ChartAnno: Evaluating MLLMs for Chart Annotation Generation 提出ChartAnno基准以评估多模态大语言模型的图表注释生成能力 large language model multimodal
20 DocTrace: Towards Traceable Long Document VQA via Hierarchical Evidence Graph Reasoning 提出DocTrace以解决长文档VQA的可追溯性问题 large language model multimodal
21 ContinualSkillBench: Can LLM Agents Truly Evolve Their Capabilities? 提出ContinualSkillBench以评估LLM代理的技能演化能力 large language model
22 ReflectRL: Learning from Golden Negative Trajectories via Reflective-to-Direct Reasoning 提出ReflectRL以利用黄金负轨迹提升推理能力 large language model
23 The Transformer Revolution, Part 1: Dynamic Processing through Output- Weight Interconnections 提出SIDPP以解决Transformer推理中的动态处理问题 large language model
24 Intertemporal Preference Steering in Qwen3 via Contrastive Activation Addition 通过对比激活添加实现Qwen3的时间偏好引导 large language model
25 Autoreflection: How Agentic Strange Loops Turn Human Culture into AI Infrastructure 提出自反机制以提升AI代理的文化理解能力 TAMP
26 Evaluating LLMs in Database Scenarios: A Lifecycle Benchmark for Assessing Their Potential in Core Database Tasks 提出DBLifeBench以评估LLMs在数据库生命周期中的潜力 large language model
27 Risky Business: Measuring The Faithfulness-Safety Tension 提出HazMart以解决大型推理模型的安全性与可信性矛盾问题 chain-of-thought
28 AI-Based Sound Effect Generation: A Narrative Review of Generative Models Across Input Modalities 综述AI驱动的音效生成模型以解决多模态输入的挑战 multimodal
29 MissClick: Exploiting Digit-Serialized Coordinates to Attack GUI Grounding Models 提出MissClick以攻击GUI视觉定位模型的安全漏洞 visual grounding
30 LiveEvalBench: Toward Open-World Evaluation for Web Generation 提出LiveEvalBench以解决前端生成评估的动态性问题 large language model ✅
31 AutoSND: From Execution Evidence to Structural Policies for Automated Network Dismantling Heuristic Discovery 提出AutoSND以解决网络拆解启发式设计的效率与效果问题 large language model ✅
32 MuEvo: LLM-Driven Evolution of Multi-Heuristic Ensemble 提出MuEvo框架以解决多启发式优化问题 large language model
33 Formal Verification of Agentic Systems over Operational Data 提出STEADs框架以解决LLM驱动系统的验证问题 large language model
34 FraQ: Efficient Coordinate-Space Recompression for Federated Low-Rank Adaptation 提出FraQ以解决联邦低秩适应中的聚合不匹配问题 large language model
35 Dr. AGENTONOMICS: A Didactic Experiment of AGENTONOMICS 提出AGENTONOMICS框架以提升AI代理的教学与管理能力 multimodal
36 DataSpace: Benchmarking Data Agents for Verifiable Analytics over Heterogeneous Workspaces 提出DataSpace以解决异构工作空间中的数据代理分析问题 multimodal
37 Towards Improving Sequential Decision-Making in LLM Agents via Experience Memory 通过经验记忆提升LLM代理的顺序决策能力 large language model
38 The Evolutionary Origin of Values: implications for AI alignment, sentience and existential risk 探讨生物价值起源以解决AI对齐与存在风险问题 large language model
39 Route-Align-Verify for Functional Correctness in Code Generation 提出RAV框架以提升代码生成的功能正确性 large language model
40 Evaluating LLM Trade-offs for Enterprise Automation: Lessons from Workflow Generation in a Production Enterprise Platform 提出基于大语言模型的工作流生成以应对企业合规管理挑战 large language model
41 AgentPanel: Toward a New Paradigm for Human--AI Collaboration in Exploring Scientific Questions 提出AgentPanel以解决科学探索中的人机协作问题 large language model
42 TaskPress: Query-Agnostic KV Cache Compression via Task-Guided Pruning 提出TaskPress以解决长上下文推理中的KV缓存压缩问题 large language model
43 The Agent Operating System (AOS): A Reference Operating Architecture for Distributed Agentic Systems 提出Agent操作系统以解决分布式智能系统架构问题 large language model
44 EduClaw-Bench: A Long-Horizon Benchmark for Pedagogical LLM Agents with Simulated Learners 提出EduClaw-Bench以评估长期教育代理的有效性 large language model
45 Beyond Average Performance: Dynamic Instance Clustering and Specialized Algorithm Design in LLM-Assisted Evolutionary Search 提出DyCA以解决现有LES方法的尾部鲁棒性不足问题 large language model
46 Getting the Parameters Right: A Difficulty-Graded Benchmark and Probe-Guided Training for LLM Tool Calls 提出探针引导训练框架以提升LLM工具调用参数生成准确性 large language model

🔬 支柱二：RL算法与架构 (RL & Architecture) (14 篇)

#题目一句话要点标签🔗⭐
47 FOUND-AF: Benchmarking ECG Foundation Models for Atrial Fibrillation Detection 提出FOUND-AF框架以评估心房颤动检测的ECG基础模型 JEPA foundation model
48 Rethinking Modality Reliability in Multimodal Sentiment Analysis with Incomplete Observations 提出MRCF框架以解决多模态情感分析中的模态可靠性问题 representation learning multimodal
49 Verifiable Memory: Learning Unified Memory Management with Local and Global Verifiers for Large Language Model Agents 提出可验证内存框架以优化大语言模型的记忆管理 reinforcement learning large language model ✅
50 PhyAI: Real-Time Physical AI at the Edge, Scalable Rollouts in the Cloud 提出PhyAI以统一物理AI推理流程，提高实时性能 reinforcement learning vision-language-action VLA ✅
51 Hybrid LLM-Augmented Reinforcement Learning Agents for Complex Sequential Decision Tasks 提出LLM增强的强化学习代理以解决复杂序列决策任务 reinforcement learning large language model
52 LLM Serving in the Wild: An Empirical Study of Frameworks, Methods, and System Designs 研究LLM服务框架以优化软件工程中的推理效率 reinforcement learning large language model multimodal
53 When Teachers Mislead: Spurious-Signal-Aware On-Policy Distillation 提出SA-OPD以解决语言模型中的虚假信号问题 distillation large language model
54 Neurosymbolic Reasoning with Incremental Knowledge for Sample Efficient Hierarchical Reinforcement Learning 提出增量知识的神经符号层次强化学习以解决样本效率问题 reinforcement learning reward shaping ✅
55 Enactive Artificial Intelligence: A Decision-Centric Architecture for Complex Systems 提出Enactive AI框架以解决复杂系统中的决策支持问题 world model world models large language model
56 GROW: Group-Relative Advantage-Weighted On-Policy Reinforcement Learning of Autoregressive-Diffusion Text-to-Speech model 提出GROW方法以解决流匹配文本到语音的强化学习挑战 reinforcement learning flow matching
57 TACT: Taxonomy-Aligned Post-Training for Pedagogically Adaptive English Tutoring 提出TACT框架以提升英语作为第二语言的适应性辅导效果 teacher-student large language model
58 ToolLIFT: Lifting Tool-Specific Trajectories into Function-Level Graphs for Generalizable Tool Planning 提出ToolLIFT以解决工具规划中的通用性问题 reinforcement learning large language model
59 CastFSR: A Fast--Slow--Reflect Agentic Reasoning Framework for Context-Aware Time Series Forecasting 提出CastFSR框架以解决时间序列预测中的上下文感知问题 reinforcement learning large language model ✅
60 UniNav: A Unified World-Action Diffusion Model for Visual Navigation 提出UniNav以解决视觉导航中的未来预测与动作生成问题 world model world models

🔬 支柱一：机器人控制 (Robot Control) (2 篇)

#题目一句话要点标签🔗⭐
61 When Outputs Disperse, Does Epistemic Revision Follow? A Black-Box Coupling Diagnostic for Machine Collectives 提出黑箱耦合诊断方法以解决集体智能中的认知修正问题 MPC
62 Adversarial Stress Testing of Role-Playing Language Agents using Multi-Agent Evaluation 提出多代理评估平台以应对角色扮演语言代理的对抗性压力测试问题 manipulation

⬅️ 返回 cs.AI 首页 · 🏠 返回主页