cs.CL（2025-09-18）

📊 共 46 篇论文 | 🔗 10 篇有代码

🎯 兴趣领域导航

支柱九：具身大模型 (Embodied Foundation Models) (41 🔗9) 支柱二：RL算法与架构 (RL & Architecture) (4) 支柱一：机器人控制 (Robot Control) (1 🔗1)

🔬 支柱九：具身大模型 (Embodied Foundation Models) (41 篇)

#题目一句话要点标签🔗⭐
1 Beyond Spurious Signals: Debiasing Multimodal Large Language Models via Counterfactual Inference and Adaptive Expert Routing 提出基于因果推断和自适应专家路由的多模态大语言模型去偏框架，提升复杂推理任务的鲁棒性。 large language model multimodal
2 Evaluating Multimodal Large Language Models on Spoken Sarcasm Understanding 评估多模态大语言模型在口语讽刺理解中的表现 large language model multimodal
3 Red Teaming Multimodal Language Models: Evaluating Harm Across Prompt Modalities and Models 红队测试揭示多模态大语言模型在不同模态提示下的安全漏洞 large language model multimodal
4 Decoupled Proxy Alignment: Mitigating Language Prior Conflict for Multimodal Alignment in MLLM 提出解耦代理对齐(DPA)方法，缓解MLLM中语言先验冲突，提升视觉-语言对齐性能。 large language model multimodal ✅
5 Quantifying Uncertainty in Natural Language Explanations of Large Language Models for Question Answering 针对大语言模型问答解释，提出一种自然语言解释不确定性量化框架 large language model
6 Quantifying Self-Awareness of Knowledge in Large Language Models 提出AQE方法以量化大语言模型知识自感知中的问题侧影响，并提出SCAO方法增强模型侧信号。 large language model
7 LNE-Blocking: An Efficient Framework for Contamination Mitigation Evaluation on Large Language Models 提出LNE-Blocking框架，有效评估大语言模型在潜在数据泄露下的真实性能。 large language model ✅
8 Assessing Historical Structural Oppression Worldwide via Rule-Guided Prompting of Large Language Models 提出一种基于规则引导的大语言模型框架，用于评估全球历史结构性压迫。 large language model ✅
9 What's the Best Way to Retrieve Slides? A Comparative Study of Multimodal, Caption-Based, and Hybrid Retrieval Techniques 对比多模态、文本和混合检索技术，探究幻灯片最佳检索方案 multimodal
10 Fair-GPTQ: Bias-Aware Quantization for Large Language Models Fair-GPTQ：面向大语言模型的偏见感知量化方法，提升公平性并保持性能。 large language model
11 SMARTER: A Data-efficient Framework to Improve Toxicity Detection with Explanation via Self-augmenting Large Language Models SMARTER：利用自增强大语言模型，高效提升毒性检测能力并提供可解释性 large language model
12 Large Language Model probabilities cannot distinguish between possible and impossible language 大型语言模型无法通过概率区分语法上可能和不可能的语言 large language model
13 LLM-OREF: An Open Relation Extraction Framework Based on Large Language Models 提出基于大语言模型的开放关系抽取框架LLM-OREF，无需人工干预即可泛化到新关系。 large language model ✅
14 Semantic Representation Attack against Aligned Large Language Models 提出语义表征攻击，提升大语言模型对抗攻击的成功率和自然性。 large language model
15 CLEAR: A Comprehensive Linguistic Evaluation of Argument Rewriting by Large Language Models 提出CLEAR评估框架，用于全面评估大型语言模型在论证改写任务中的语言能力。 large language model
16 A Comparative Evaluation of Large Language Models for Persian Sentiment Analysis and Emotion Detection in Social Media Texts 对比评估大型语言模型在波斯语社交媒体文本情感分析和情绪检测中的性能 large language model
17 Evaluating Large Language Models for Cross-Lingual Retrieval 评估大语言模型在跨语言检索中的应用，揭示检索器与重排序器间的相互作用。 large language model
18 MUSE: MCTS-Driven Red Teaming Framework for Enhanced Multi-Turn Dialogue Safety in Large Language Models MUSE：基于MCTS的大语言模型多轮对话安全红队测试框架 large language model ✅
19 ParlAI Vote: A Web Platform for Analyzing Gender and Political Bias in Large Language Models 提出ParlAI Vote平台以分析大型语言模型中的性别与政治偏见 large language model
20 Position: Thematic Analysis of Unstructured Clinical Transcripts with Large Language Models 利用大型语言模型进行非结构化临床记录的主题分析，并提出标准化评估框架。 large language model
21 TriSPrompt: A Hierarchical Soft Prompt Model for Multimodal Rumor Detection with Incomplete Modalities 提出TriSPrompt以解决多模态数据中谣言检测的缺失模态问题 multimodal
22 GeoResponder: Towards Building Geospatial LLMs for Time-Critical Disaster Response GeoResponder：构建用于时间敏感型灾害响应的地理空间大语言模型 large language model foundation model
23 UnifiedVisual: A Framework for Constructing Unified Vision-Language Datasets 提出UnifiedVisual框架，构建统一视觉语言数据集，促进多模态理解与生成协同。 large language model multimodal ✅
24 TableDART: Dynamic Adaptive Multi-Modal Routing for Table Understanding TableDART：动态自适应多模态路由用于表格理解 large language model multimodal
25 An Evaluation-Centric Paradigm for Scientific Visualization Agents 提出科学可视化Agent的评测范式，促进Agent能力提升与领域创新 large language model
26 LLM-Assisted Topic Reduction for BERTopic on Social Media Data 提出LLM辅助的BERTopic主题降维方法，提升社交媒体数据主题建模效果 large language model
27 PILOT: Steering Synthetic Data Generation with Psychological & Linguistic Output Targeting PILOT：利用心理语言学输出目标引导合成数据生成，提升控制精度。 large language model
28 Real, Fake, or Manipulated? Detecting Machine-Influenced Text 提出HERO模型，用于区分人类撰写、机器生成、机器润色和机器翻译的文本。 large language model
29 PolBiX: Detecting LLMs' Political Bias in Fact-Checking through X-phemisms PolBiX：通过委婉语检测大型语言模型在事实核查中的政治偏见 large language model
30 ATTS: Asynchronous Test-Time Scaling via Conformal Prediction ATTS：通过保形预测实现异步测试时扩展，显著加速大语言模型推理。 large language model ✅
31 TextMineX: Data, Evaluation Framework and Ontology-guided LLM Pipeline for Humanitarian Mine Action TextMineX：构建人道主义排雷领域知识抽取数据集与本体引导的LLM流程 large language model
32 Benchmarking and Improving LLM Robustness for Personalized Generation 提出PERG框架与Pref-Aligner方法，提升LLM在个性化生成中的事实性与鲁棒性。 large language model
33 SynParaSpeech: Automated Synthesis of Paralinguistic Datasets for Speech Generation and Understanding SynParaSpeech：自动化合成口语数据集，提升语音生成与理解 TAMP ✅
34 Explicit vs. Implicit Biographies: Evaluating and Adapting LLM Information Extraction on Wikidata-Derived Texts 通过LoRA微调提升LLM在Wikidata文本信息抽取中处理隐式信息的能力 large language model
35 LLM Agents at the Roundtable: A Multi-Perspective and Dialectical Reasoning Framework for Essay Scoring 提出Roundtable Essay Scoring (RES)框架，利用多智能体辩证推理提升作文自动评分效果。 large language model
36 ReCoVeR the Target Language: Language Steering without Sacrificing Task Performance 提出ReCoVeR，通过语言引导向量减少LLM的语言混淆，同时保持任务性能。 large language model ✅
37 Reasoning over Boundaries: Enhancing Specification Alignment via Test-time Deliberation 提出Align3，通过测试时审议增强LLM在动态规范下的对齐能力 large language model
38 Reveal and Release: Iterative LLM Unlearning with Self-generated Data 提出Reveal-and-Release迭代框架，利用自生成数据实现大语言模型高效遗忘 large language model
39 Controlling Language Difficulty in Dialogues with Linguistic Features 提出基于语言特征控制的对话系统，提升语言学习交互体验 large language model
40 Catch Me If You Can? Not Yet: LLMs Still Struggle to Imitate the Implicit Writing Styles of Everyday Authors 评估大语言模型模仿个人写作风格能力：现有模型在非正式文体中表现不足 large language model
41 Introducing OmniGEC: A Silver Multilingual Dataset for Grammatical Error Correction OmniGEC：提出多语言语法纠错的银标准数据集，促进跨语言GEC模型发展 large language model

🔬 支柱二：RL算法与架构 (RL & Architecture) (4 篇)

#题目一句话要点标签🔗⭐
42 Cross-Modal Knowledge Distillation for Speech Large Language Models 提出跨模态知识蒸馏框架，解决语音大语言模型中的灾难性遗忘和模态不等价问题 distillation large language model
43 Delta Knowledge Distillation for Large Language Models 提出Delta-KD，通过保留分布偏移量提升大语言模型知识蒸馏效果 distillation large language model
44 HARNESS: Lightweight Distilled Arabic Speech Foundation Models 提出HArnESS：轻量级阿拉伯语语音基础模型，适用于资源受限环境 distillation foundation model
45 Empathy-R1: A Chain-of-Empathy and Reinforcement Learning Framework for Long-Form Mental Health Support 提出Empathy-R1框架，结合同理心链式推理与强化学习，提升长文本心理健康支持质量。 reinforcement 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🔗⭐
46 V-SEAM: Visual Semantic Editing and Attention Modulating for Causal Interpretability of Vision-Language Models 提出V-SEAM以解决视觉语言模型的因果可解释性问题 manipulation multimodal ✅

⬅️ 返回 cs.CL 首页 · 🏠 返回主页