cs.AI（2025-09-03）

📊 共 10 篇论文 | 🔗 1 篇有代码

🎯 兴趣领域导航

#	题目	一句话要点	标签	🔗
1	ANNIE: Be Careful of Your Robots	ANNIE：针对具身AI系统的对抗性安全攻击研究与基准测试	embodied AI vision-language-action VLA	✅
2	Designing Gaze Analytics for ELA Instruction: A User-Centered Dashboard with Conversational AI Support	设计用户中心的注视分析仪表盘以支持ELA教学	large language model multimodal
3	Are LLM Agents Behaviorally Coherent? Latent Profiles for Social Simulation	揭示LLM智能体行为不一致性，质疑其在社会模拟中替代人类受试者的能力	large language model
4	RAGuard: A Novel Approach for in-context Safe Retrieval Augmented Generation for LLMs	RAGuard：面向LLM，用于安全检索增强生成的新框架，提升海上风电维护安全性。	large language model
5	Towards a Neurosymbolic Reasoning System Grounded in Schematic Representations	提出Embodied-LM，通过具身图式表征增强神经符号推理能力	large language model
6	Explainable Knowledge Graph Retrieval-Augmented Generation (KG-RAG) with KG-SMILE	提出KG-SMILE框架，提升知识图谱RAG的可解释性，解决幻觉问题。	large language model

#	题目	一句话要点	标签
7	Emergent Hierarchical Reasoning in LLMs through Reinforcement Learning	提出HICRA算法，通过强化学习提升LLM的层级推理能力，优化策略规划。	reinforcement learning large language model
8	The Personality Illusion: Revealing Dissociation Between Self-Reports & Behavior in LLMs	揭示LLM人格幻觉：自述与行为之间的解离现象	RLHF large language model
9	CausalARC: Abstract Reasoning with Causal World Models	提出CausalARC，用于在低数据和分布偏移下进行因果抽象推理的实验平台。	world model
10	Language Models Do Not Follow Occam's Razor: A Benchmark for Inductive and Abductive Reasoning	提出InAbHyD基准测试LLM的归纳和溯因推理能力，发现其不遵循奥卡姆剃刀原则	world model large language model