About Me

• I am an Assistant Professor in the School of Computer Engineering and Science, Shanghai University. I received my Ph.D. degree from Shanghai Jiao Tong University, advised by Prof. Gongshen Liu and Prof. Zhuosheng Zhang.
• My research interests include Large Language Models (LLMs), AI Agents, Multimodal AI, AI Security (Backdoor Attacks, Watermarking, Privacy), and Cybersecurity.
• I have published 24+ papers in top-tier venues, including ICML, ACL, AAAI, NAACL, TNNLS, and others. I serve as a reviewer for leading conferences and journals, including NeurIPS, ICLR, ACL, AAAI, TITS, and TVT.
• 🎓 Prospective Students & Collaborators
  If you are interested in my research, feel free to email me or contact me via WeChat. I am always happy to discuss research ideas, explore collaborations, or build exciting AI systems together.

  ---

  🌱 欢迎加入我们！
  研究不应该是一个人的战斗，而是一群人共同探索未知的旅程。 如果你对 大模型、智能体、 多模态 AI、人工智能安全 等方向感兴趣，欢迎通过 邮件或微信联系我。 无论是讨论科研想法、合作研究，还是一起"造轮子"，我都非常期待。
  
  📌 特别欢迎计划于 2027 年选择我作为导师的本科生和硕士生！

News

Affiliation

• AI Security Lab
Shanghai Jiao Tong University

Honors and Awards

• First-class scholarship (3 times), Second-class scholarship (1 time), Blue Bridge Cup Competition Scholarship (2 times). 2015-2019
• First-class scholarship (2 times), second-class scholarship (1 time), mathematical modeling award (3 times), school-level outstanding Graduate model, school-level outstanding graduate student. 2019-2022
• National Scholarship, Principal Medal. 2021
• Cybersecurity Scholarship (Shutang), Bode Scholarship (TikTok), Outstanding Communist Youth League Member. 2023-2026

Tutorials and Contributions

• Participating: 《大模型开发全流程》Course Series Course Link
  • 基于昇腾平台搭建从零开始大模型课程学习 (主讲大模型本地部署、服务化、评估、RAG)
  • 实验手册、实验代码全部开源 (5K)

Programs

• Participating and Applying: 《面向网络暴力治理的群体行为深度感知溯源与处置技术研究》(纵向课题，国家重点研发，子课题撰写)
  • 研发多模态内容倾向性分析技术，精准识别网络暴力苗头，实现风险早发现、早干预
  • 构建演化路径感知技术，全要素挖掘网暴事件传播链条，快速锁定恶意推手与关键节点
  • 开发群体情绪极化引导技术，动态平衡公众认知，有效化解网络暴力引发的情绪失控问题
• Participating: 《面向大模型智能体应用的任务安全用例自动化构造技术和任务安全判定技术》(横向课题，项目技术参与)
  • 面向大模型智能体应用的安全探测和判定
  • 实现多模态模型赋能的智能代理环境注入攻击
• Organization and Participating: 《多模态大模型图形化用户界面推理技术》(横向课题， 项目技术Leader)
  • 攻克环境感知、决策规划，反思推理等核心挑战
  • 实现多模态模型赋能的智能代理“可见可说可执行”
• Organization and Participating: 《试油关键参数大数据分析及预测技术研究》(横向课题，中石化，项目技术完成及结项报告)
  • 实现试油关键参数的特征重要性分析和选择
  • 基于深度学习实现试油的产量预测
• Organization and Participating: 《智能网联汽车网联架构及其信息安全防护技术研究》(纵向课题，省部级项目)
  • ECU之间的零知识证明
  • 基于深度学习的车载入侵检测

Selected Publications

Thanks to all collaborators.

• TrojanRAG: Retrieval-augmented generation can be backdoor driver in large language models
  Pengzhou Cheng, Yidong Ding, Tianjie Ju, Zongru Wu, Wei Du, Ping Yi, Zhuosheng Zhang^†, Gongshen Liu^†
  Neural Networks, Under Review, CCF-B

• Transferring Backdoors between Large Language Models by Knowledge Distillation
  Pengzhou Cheng, Zongru Wu, Tianjie Ju, Wei Du, Zhuosheng Zhang^†, Gongshen Liu^†
  Under Review

• Smoothing Grounding and Reasoning for MLLM-Powered GUI Agents with Query-Oriented Pivot Tasks
  Zongru Wu, Pengzhou Cheng, Zheng Wu, Tianjie Ju, Zhuosheng Zhang^†, Gongshen Liu^†
  ACMMM 2025, Under Reveiw, CCF-A

• Investigating the Adaptive Robustness with Knowledge Conflicts in LLM-based Multi-Agent Systems
  Tianjie Ju, Bowen Wang, Hao Fei, Mong-Li Lee, Wynne Hsu, Yun Li, Qianren Wang, Pengzhou Cheng, Zongru Wu, Zhuosheng Zhang^†, Gongshen Liu^†
  Neurips 2025, Under Reveiw, CCF-A

• Atomic-to-Compositional Generalization for Mobile Agents with A New Benchmark and Scheduling System
  Yuan Guo, Tingjia Miao, Zheng Wu, Pengzhou Cheng, Ming Zhou, Zhuosheng Zhang^†
  Neurips 2025, Under Reveiw, CCF-A

• VeriOS: Query-Driven Proactive Human-Agent-GUI Interaction for Trustworthy OS Agents
  Zheng Wu, Heyuan Huang, Xingyu Lou, Xiangmou Qu, Pengzhou Cheng, Zongru Wu, Weiwen Liu, Weinan Zhang, Jun Wang, Zhaoxiang Wang, Zhuosheng Zhang^†
  CHI 2026, Under Reveiw, CCF-A

• Mobile-Aptus: Confidence-Driven Proactive and Robust Interaction in MLLM-based Mobile-Using Agents
  Zheng Wu^*, Pengzhou Cheng ^*, Zongru Wu, Yuan Guo, Gongshen Liu^†, Zhuosheng Zhang^†
  IEEE Transactions on Audio, Speech and Language Processing (JCR-Q1, IF=5.0, CCF-B), 2026

• Faithful Mobile GUI Agents with Guided Advantage Estimator
  Haowen Hu^*, Pengzhou Cheng^*, Zheng Wu, Lingzhong Dong, Gongshen Liu^†, Zhuosheng Zhang^†
  ICML 2026, CCF-A

• See, Think, Act: Teaching Multimodal Agents to Effectively Interact with GUI by Identifying Toggles
  Zongru Wu, Rui Mao, Zhiyuan Tian, Pengzhou Cheng, Tianjie Ju, Zheng Wu, Lingzhong Dong, Haiyue Sheng, Zhuosheng Zhang^†, Gongshen Liu^†
  CVPR 2026, CCF-A

• Domain Adaptation of MLLM-based Computer-Using Agents with Standard Operating Procedure
  Lingzhong Dong, Ziqi Zhou, Pengzhou Cheng, Zeng Wu, Zongru Wu, Gongshen Liu^†, Zhuosheng Zhang^†
  KESM 2026, CCF-C

• GEM: Gaussian Embedding Modeling for Out-of-Distribution Detection in GUI Agents
  Zheng Wu, Pengzhou Cheng, Zongru Wu, Lingzhong Dong, Zhuosheng Zhang^†
  AAAI 2026, CCF-A

• Flooding spread of manipulated knowledge in llm-based multi-agent communities
  Tianjie Ju, Yiting Wang, Xinbei Ma, Pengzhou Cheng, Haodong Zhao, Yulong Wang, Lifeng Liu, Jian Xie, Zhuosheng Zhang^†, Gongshen Liu^†
  Science China Information Sciences, 2025, CCF-A

• Hidden Ghost Hand: Unveiling Backdoor Vulnerabilities in MLLM-Powered Mobile GUI Agents
  Pengzhou Chengstrong>, Haowen Hu, Zheng Wu, Zongru Wu, Tianji Ju, Zhuosheng Zhang^†, Gongshen Liu^†
  EMNLP findings, 2025, CCF-B

• OS-Kairos: Adaptive Interaction for MLLM-Powered GUI Agents
  Pengzhou Cheng, Zheng Wu, Zongru Wu, Aston Zhang, Zhuosheng Zhang^†, Gongshen Liu^†
  ACL findings, 2025, CCF-A

• MKF-ADS: Multi-Knowledge Fusion Based Self-supervised Anomaly Detection System for Control Area Network
  Pengzhou Cheng, Shouxuan Liu, Zongru Wu, Lixing Tan, Gongshen Liu^†
  IEEE Transactions on Vehicular Technology (JCR-Q1, IF=6.8, CCF-B), 2025

• Deep-Learning Integrated Bioelectronic-Tissue Interface for Cardiovascular Diagnosis and Prognosis
  Shuaimin Tang, Pengzhou Cheng (Equal Contribution), Hanzhang Liang, Dongyong Sha, Lingfei Zhao, Ding Ding, Gongshen Liu, Changsheng Liu, Yiwei Liu^†, Yifan Ma^†, and Yuan Yuan^†.
  Advanced Functional Materials, (JCR-Q1, IF=19.6), 2025

• Backdoor Attacks and Countermeasures in Natural Language Processing Models: A Comprehensive Security Review
  Pengzhou Cheng, Zongru Wu, Wei Du, Haodong Zhao, Wei Lu, Gongshen Liu^†
  IEEE Transactions on Neural Networks and Learning Systems (JCR-Q1, IF=10.8, CCF-B), 2025

• SynGhost: Invisible and Universal Task-agnostic Backdoor Attack via Syntactic Transfer
  Pengzhou Cheng, Wei Du, Zongru Wu, Fengwei Zhang, Libo Chen, Zhuosheng Zhang^†, Gongshen Liu^†
  NAACL findings, 2025, CCF-B

• LSF-IDM: Deep learning-based lightweight semantic fusion intrusion detection model for automotive
  Pengzhou Cheng, Lei Hua, Haobin Jiang, Gongshen Liu^†
  Peer-to-Peer Networking and Applications (JCR-Q1, IF=3.3, CCF-C).

• 面向语言模型的文本后门防御综述
  Zongru Wu, Pengzhou Cheng, Zhuosheng Zhang, Gongshen Liu^†
  信息安全学报 (CCF-B, IF=2.16)

• Gracefully Filtering Backdoor Samples for Generative Large Language Models without Retraining
  Zongru Wu, Pengzhou Cheng, Lingyong Fang, Zhuosheng Zhang^†, Gongshen Liu^†
  COLING 2025, CCF-B

• Acquiring clean language models from backdoor poisoned datasets by downscaling frequency space
  Zongru Wu, Zhuosheng Zhang^†, Pengzhou Cheng, Gongshen Liu^†
  ACL 2024, CCF-A

• BackdoorMBTI: A Backdoor Learning Multimodal Benchmark Tool Kit for Backdoor Defense Evaluation
  Haiyang Yu, Tian Xie, Jiaping Gui, Pengyang Wang, Pengzhou Cheng, Ping Yi^†, Yue Wu^†
  KDD, 2025, CCF-A

• PLMmark: A Secure and Robust Black-box Watermarking Framework for Pre-trained Language Models
  Peixuan Li, Pengzhou Cheng, Fangqi Li, Wei Du, Haodong Zhao, Gongshen Liu^†
  AAAI, 2023, CCF-A

• DESC-IDS: Towards an efficient real-time automotive intrusion detection system based on deep evolving stream clustering
  Pengzhou Cheng, Mu Han^†, Gongshen Liu
  Future Generation Computer Systems, 2023, (JCR-Q1, IF=6.2, CCF-C)

• STC-IDS: Spatial–temporal correlation feature analyzing based intrusion detection system for intelligent connected vehicles
  Pengzhou Cheng, Mu Han^†, Aoxue Li, Fengwei Zhang
  International Journal of Intelligent Systems, 2022, (JCR-Q1, IF=5.9, CCF-C)

• TCAN-IDS: Intrusion Detection System for Internet of Vehicle Using Temporal Convolutional Attention Network
  Pengzhou Cheng, Kai Xu, Simin Li, Mu Han^†
  Symmetry, 2022, (JCR-Q2, IF=2.2)

• PPM-InVIDS: Privacy protection model for in-vehicle intrusion detection system based complex-valued neural network
  Mu Han (My Supervioser)^†, Pengzhou Cheng, Shidian Ma
  Vehicular Communication, 2021, (JCR-Q1, IF=8.9)

• Implementing an efficient secure attribute-based encryption system for iov using association rules
  Mu Han (My Supervisor)^†, Mengli Zhu, Pengzhou Cheng, Haixing Qu, Shidiam Ma
  Symmetry, 2022, (JCR-Q2, IF=2.2)

• Zero-knowledge identity authentication for internet of vehicles: Improvement and application
  Mu Han^†, Zhikun Yin, Pengzhou Cheng, Xing Zhang, Shidian Ma
  Plos One, 2020, (JCR-Q1, IF=2.9)

• Cvnns-ids: Complex-valued neural network based in-vehicle intrusion detection system
  Mu Han (My Supervisor)^†, Pengzhou Cheng, Shidian Ma
  Security and Privacy in Digital Economy: First International Conference, SPDE 2020, Oral

Academic Service

• Journal Reviewer: IEEE TPAMI, TDSC, TIFS, TITS, TVT, TBD, TII, TIST, TIT, TNSE, TSC, TKDD, TIOT, JBHI, IoTJ, TSUSC, Science China Information Sciences (SCIS), Journal of Supercomputing, Peer-to-Peer Network and Applications, Journal of Intelligent Systems, ACM transactions on knowledge discovery from data, Cluster computing, IET communications, Cybernetics & Systems, Unmanned Systems.
• Conference Reviewer: ACL2024, EMNLP 2024, NAACL 2025, ACL 2025, ICLR 2025, NeurIPS 2024, SIGKDD 2025, IJCNN 2025, ACMMM 2025.

Teaching Service

• Teaching Assistant for "Hands-on deep learning", Fall 2021; "Natural Language Processing" , Fall 2023; "Principles of Computer Viruses"
• Guided a graduate student to complete literature research, idea generation and submission to JCR Q1 on Internet of Vehicles security, which was successfully accepted by Peer to Peer Network and Application
• Guided a graduate student to complete literature research, idea conception and JCR Q1 submission in the field of pre-trained intrusion detection, which was successfully accepted by Journal of SuperComputing
• 6 undergraduate designations and 2 master's degree supervision

Research Team

PhD Students

• 2025: Xinjie Lai, Weihao Zuo (co-advising with Prof. Xinpeng Zhang)

Master Students

• 2026: Haowen Hu (co-advising with Prof. Zhuosheng Zhang)
• 2025: Zheng Wu (co-advising with Prof. Zhuosheng Zhang)
• 2024: Lingzhong Dong (co-advising with Prof. Zhuosheng Zhang)

Alumni

• 2024: Xiaochen Li