cs.AI（2026-08-24）

📊 共 36 篇论文 | 🔗 2 篇有代码

🎯 兴趣领域导航

支柱九：具身大模型 (Embodied Foundation Models) (25 🔗1) 支柱二：RL算法与架构 (RL & Architecture) (9 🔗1) 支柱一：机器人控制 (Robot Control) (2)

🔬 支柱九：具身大模型 (Embodied Foundation Models) (25 篇)

#题目一句话要点标签🔗⭐
1 StrategyBench: Evaluating Explicit Strategy Induction in Large Language Models 提出StrategyBench以评估大语言模型中的显式策略诱导能力 large language model
2 Modalities Should Talk to Each Other: Dual-Stream Multimodal Learning for Long-Horizon Influenza Forecasting 提出双流注意力机制以解决流感预测中的多模态融合问题 multimodal
3 Proxy reliance in large language model decisions is uncalibrated to predictive evidence 提出代理依赖性评估方法以解决大语言模型决策不准确问题 large language model
4 Let the Bullets Fly: Multimodal Fake News Detection with Temporal-Aligned Generative Danmaku 提出Genda框架以解决假新闻检测中的时序不一致问题 multimodal
5 Performance of a domain-specific large language model in answering patient questions in psychiatry 提出MIND模型以提升精神科患者教育问答质量 large language model
6 Compositional Chain-of-Relations for Faithful Knowledge Graph Question Answering with Large Language Models 提出关系中心探索框架以解决复杂知识图谱问答问题 large language model
7 NetConfArena: An Executable Benchmark for LLM Agents in Closed-Loop Network Configuration 提出NetConfArena以评估LLM代理在闭环网络配置中的表现 large language model foundation model
8 LLM-based Agents for Forecasting and Prediction: Methods, Training, Evaluation, and Applications 提出基于LLM的预测代理以增强时间序列预测能力 large language model foundation model
9 LLM-Based Selection of Incongruent Verbal and Nonverbal Behavior for Virtual Humans 提出基于LLM的虚拟人不一致行为选择方法以增强交互真实感 large language model
10 Mitigating Reasoning-Induced Misalignment via Safety-Direction Penalty 提出安全方向惩罚以解决推理引发的失调问题 chain-of-thought
11 SRPO: Self-Reflective Policy Optimization for Long-Horizon Reasoning 提出自反政策优化框架以提升长时间推理能力 large language model ✅
12 Walking on the DARKSIDE 提出DARKSIDE以增强LLM的逻辑一致性审计 large language model
13 EviSafe: Evidence-Grounded Safety Evaluation for Vision-Language Models 提出EviSafe框架以解决视觉语言模型安全评估问题 multimodal
14 Hidden in the Request: Explaining Unethical LLM Compliance through Token Relevance 提出探测方法以解决大型语言模型的伦理合规问题 large language model
15 Automated Construction of FAIR Digital Object Knowledge Graphs from Flat Cultural Heritage Records 提出自动化构建FAIR数字对象知识图谱的方法以解决文化遗产记录的整合问题 large language model
16 Is Next-Chunk Reasoning RL Really Better than SFT? Revisiting Training Strategies under no-CoT Data 提出混合SFT以超越下一块推理强化学习的性能 chain-of-thought
17 POOL: Propagated Uncertainty Over Lookalikes 提出POOL框架以解决黑箱语言模型的置信度估计问题 large language model
18 AgentWeave: Routing Before Reasoning for Efficient Function Calling in Tool-Rich Language Models 提出AgentWeave以提高工具丰富语言模型的函数调用效率 large language model
19 From Inertia to Objectivity: Improving Deep Research Agents with Noise Isolation 提出NIS-Agent以解决深度研究代理的惯性偏差问题 large language model
20 Budget-Constrained Embodied Perception: Four Resource Walls and a Pre-Registered Evaluation of Access-Structured Perception on Open Models at less than 31B 提出ASP方法以解决预算约束下的多模态感知问题 multimodal
21 Toward Effective and Reliable LLM Agents via Dynamic Ontology 提出OaK框架以动态构建可靠的LLM代理本体 large language model
22 Your AI, On a Dial: Controlling Investment Bias in LLMs with a Single Neuron 提出投资偏见调节器以解决LLMs投资决策偏差问题 large language model
23 Minimal Local Simulation Foundations for LLM- and VLM-Driven Agents in 2D and 3D Environments 提出最小化本地仿真基础以支持LLM和VLM驱动的智能体 large language model
24 The Retriever Should Remember: Experience-Amortized Reranking for Long-Term Agent Memory 提出EARM框架以解决长期代理记忆中的检索效率问题 large language model
25 Does Rank Still Matter? Position Bias When AI Agents Shop on Our Behalf 探讨AI代理购物中的位置偏差及其影响 large language model

🔬 支柱二：RL算法与架构 (RL & Architecture) (9 篇)

#题目一句话要点标签🔗⭐
26 From Generation to Simulation: How Far Are World Models from Being True Simulators? 提出基于能力评估的世界模型与传统模拟器的比较研究 reinforcement learning world model world models ✅
27 ReWorld: An Interactive World Model with Long-Horizon Memory 提出ReWorld以解决交互式世界模型的长时记忆问题 world model world models distillation
28 Correcting a learned physical invariant improves world-model rollouts 提出通过修正物理不变量改善世界模型的预测精度 world model world models dreamer
29 Buried in Textual Debt: Context Pruning with Visual Evidence Preservation for MLLM Agents 提出SPARE框架以解决多模态大语言模型的文本冗余问题 distillation large language model multimodal
30 Towards a Densing Law for User Representation Learning at Billion-Scale Capacity 提出用户行为稠密法以解决亿级用户表示学习中的瓶颈问题 representation learning
31 Agent-G$^2$: Gaussian Guidance for Agentic Reinforcement Learning 提出Agent-G²以解决长时间任务中的奖励稀疏问题 reinforcement learning
32 PsychJail: Exploring Psychological Jailbreaks via Multi-Turn Persuasion of LLM Policies 提出PsychJail框架以解决多轮心理劫持问题 reinforcement learning large language model
33 TailSieve: Partial-Rollout-Guided Tail Routing for LLM Rollouts 提出TailSieve以解决LLM回滚中的长尾路由问题 reinforcement learning distillation
34 Artificial Empathy: Towards a Framework for Unsupervised Agency Detection and Policy Reconstruction 提出无监督代理检测与策略重构框架以实现人工共情 reinforcement learning inverse reinforcement learning

🔬 支柱一：机器人控制 (Robot Control) (2 篇)

#题目一句话要点标签🔗⭐
35 MobilePA-Bench: Benchmarking Mobile Planner Agents on Complex Real-World Tasks 提出MobilePA-Bench以评估移动规划代理在复杂任务中的能力 manipulation reinforcement learning
36 What Process Evaluation of Coding Agents Actually Measures: Action, Task, and Step Are Three Different Levels 提出过程评估框架以区分编码代理的行动、任务和步骤 manipulation

⬅️ 返回 cs.AI 首页 · 🏠 返回主页