姓名：刘燕兵 正高级工程师，博士生导师
工作单位：中国科学院信息工程研究所，信息内容安全国家工程研究中心
联系地址：北京市海淀区树村路19号，邮编：100193
电子邮件：liuyanbing@iie.ac.cn
简介：刘燕兵，中国科学院信息工程研究所正高级工程师、博士生导师，信息内容分析研究室主任，中国科学院大学网络空间安全学院岗位教授。主要开展多模态信息智能处理、网络数据挖掘、信息内容安全、人工智能安全等方向的研究工作，在国内外重要学术会议和期刊上发表论文130余篇，申请和授权发明专利70余项，指导博士和硕士研究生30余人，在国科大讲授《信息内容安全》核心课程。承担了科技部、中国科学院、基金委以及国家相关部门的二十余项科研任务，担任国家重点研发计划项目负责人、国家重大工程项目技术副总师，牵头研制的技术平台在国家重要部门和地区规模化应用，取得显著应用效果。获得新疆科技进步一等奖、中央办公厅科技进步二等奖、中国科学院大学优秀指导教师、CollabrateCOM最佳论文奖、CNCERT中国信息安全技术公开赛第一名等奖励和荣誉。
工作经历
2012.1至今：中国科学院信息工程研究所，历任助理研究员、副研究员、正高级工程师，博士生导师
2006.7-2011.12：中国科学院计算技术研究所，历任研究实习员、助理研究员
研究方向
多模态信息智能处理、网络数据挖掘、信息内容安全、人工智能安全
论文专利软著
1.MIRAGE: How Conversation State Shapes Historical Evidence Use in Multimodal Personal Agents, ACM International Conference on Multimedia (ACM MM), 2026
2.Music Hallucination in Audio-Language Models: A Hierarchical Formulation and Empirical Study, ACM International Conference on Multimedia (ACM MM), 2026
3.Trans-RAG: Query-Centric Vector Transformation for Secure Cross-Organizational Retrieval, International Conference on Database Systems for Advanced Applications (DASFAA), 2026
4.STIndex: A Context-Aware Multi-Dimensional Spatiotemporal Information Extraction System, WWW Companion '26: Companion Proceedings of the ACM Web Conference 2026, 2026
5.MemGovern: Enhancing Code Agents through Learning from Governed Human Experiences, Annual Meeting of the Association for Computational Linguistics (ACL 2026 Findings), 2026
6.Conformal Event Prediction with Temporal Knowledge Graph, Annual Meeting of the Association for Computational Linguistics (ACL 2026 Findings), 2026
7.LitVISTA: A Benchmark for Narrative Orchestration in Literary Text, Annual Meeting of the Association for Computational Linguistics (ACL 2026 MainConference), 2026
8.Two Streams, One Sarcasm: Orthogonal Expert Tuning for Holistic Multimodal Sarcasm Understanding, Annual Meeting of the Association for Computational Linguistics (ACL 2026 MainConference), 2026
9.MuVaC: A Variational Causal Framework for Multimodal Sarcasm Understanding in Dialogues, The Web Conference (WWW), 2026
10.OPERA: A Reinforcement Learning-Enhanced Orchestrated Planner-Executor Architecture for Reasoning-Oriented Multi-Hop Retrieval, AAAI Conference on Artificial Intelligence (AAAI), 2026
11.Beyond Accuracy: A Cognitive Load Framework for Mapping the Capability Boundaries of Tool-use Agents, AAAI Conference on Artificial Intelligence (AAAI), 2026
12.RepGuard: Adaptive Feature Decoupling for Robust Backdoor Defense in Large Language Models, Annual Conference on Neural Information Processing Systems (NuerIPS Main Conference Track), 2025
13.Can We Steer Reasoning Direction by Thinking Intervention?, Conference on Empirical Methods in Natural Language Processing (EMNLP Findings), 2025
14.Multi-View Incongruity Learning for Multimodal Sarcasm Detection, International Conference on Computational Linguistics (COLING), 2025
15.Dialogues Aspect-based Sentiment Quadruple Extraction via Structural Entropy Minimization Partitioning, ACM International Conference on Information and Knowledge Management (CIKM), 2025
16.Emotion Transfer with Enhanced Prototype for Unseen Emotion Recognition in Conversation, Conference on Empirical Methods in Natural Language Processing (EMNLP), 2025
17.ReTD: Reconstruction-Based Traceability Detection for Generated Images, IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2025
18.Counterfactual-Augmented Representation Learning based Event Prediction, IEEE International Conference on Multimedia and Expo (ICME), 2025
19.T-T: Table Transformer for Tagging-based Aspect Sentiment Triplet Extraction, International Joint Conference on Artificial Intelligence (IJCAI), 2025
20.FairCDR: Transferring Fairness and User Preferences for Cross-Domain Recommendation, Proceedings of the 31st ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD), 2025
21.RDGCN: Reinforced Dependency Graph Convolutional Network for Aspect-based Sentiment Analysis, ACM International Conference on Web Search and Data Mining (WSDM), 2024
22.Mulan: A Multi-Level Alignment Model for Video Question Answering, Conference on Empirical Methods in Natural Language Processing (EMNLP Findings), 2023