Fei Zhu

Assistant Professor [Google Scholar] Centre for Artificial Intelligence and Robotics, Hong Kong Institute of Science & Innovation Chinese Academy of Sciences 3/F, 17W, Science Park West Avenue, Hong Kong Science Park, Hong Kong Email: zhfei2018@gmail.com WeChat: 17888841931

About Me

I am currently an Assistant Professor at the Centre for Artificial Intelligence and Robotics, Hong Kong Institute of Science & Innovation, Chinese Academy of Sciences, working with Prof. Zhaoxiang Zhang and Prof. Gaofeng Meng. I received my Ph.D. in Pattern Recognition and Intelligent Systems from the Institute of Automation, Chinese Academy of Sciences, where I was advised by Prof. Cheng-Lin Liu and Prof. Xu-Yao Zhang. Prior to this, I received the B.E. degree from Tsinghua University.

Research Highlights: My research focuses on both the theoretical and applied aspects of dynamic learning, especially for foundation models such as MLLMs and LLMs. Additionally, I am keen on utilizing these models to facilitate applications in biomedicine and embodied robotics. 

Related ML topics: continual pre-training, continual post-training, reinforcement fine-tuning, AI alignment.
Focused applications: biomedicine and healthcare, robot learning and embodied AI. 

We are looking for collaborators, who are self-motivated and have a solid foundation in mathematics and programming. If you are interested, please contact me via email (zhfei2018@gmail.com) or WeChat (17888841931).

Invited Talks

• Unknown Rejection in Open Environment, Biomedical Engineering Distinguished Lecture Series, 南方科技大学, August, 2024

• Deep Continual Learning, School of Computer Science and Engineering, 南京理工大学, January, 2025

• Open-Environment Continual Learning, 中关村人工智能研究院, 北京, February, 2025

• Continual Learning in Multimodal Large Language Model, VALSE 2025 持续学习论坛, 珠海, June, 2025

• Continual Learning: Theory, Methods and Applications, 2025 中国图象图形学学会青年科学家会议, 青岛, September, 2025

• Recent Advance of Continual Learning, 北京大学深圳研究生院, October, 2025

Selected Publications

• [NeurIPS 2025 Spotlight Paper] RobustMerge: Parameter-Efficient Model Merging for MLLMs with Direction Robustness [paper].
  Fanhu Zeng, Haiyang Guo, Fei Zhu^📧, Li Shen, Hao Tang.
  

• [NeurIPS 2025] C-NAV: Towards Self-Evolving Continual Object Navigation in Open World [paper].
  Mingming Yu, Fei Zhu, Wenzhuo Liu, Yirong Yang, Yunbo Wang, Wenjun Wu, Jing Liu.
  

• [TPAMI 2025] PASS++: A Dual Bias Reduction Framework for Non-Exemplar Class-Incremental Learning [paper].
  Fei Zhu, Xu-Yao Zhang, Zhen Cheng, Cheng-Lin Liu.
  

• [TPAMI 2024] Revisiting Confidence Estimation: Towards Reliable Failure Prediction [paper] [arxiv] [code].
  Fei Zhu, Xu-Yao Zhang, Zhen Cheng, Cheng-Lin Liu.
  

• [TPAMI 2023] Learning by Seeing More Classes [paper].
  Fei Zhu, Xu-Yao Zhang, Rui-Qi Wang, Cheng-Lin Liu.
  

• [Neural Networks 2023] Imitating the Oracle: Towards Calibrated Model for Class Incremental Learning [paper] [code].
  Fei Zhu, Zhen Cheng, Xu-Yao Zhang, Cheng-Lin Liu.
  

• [CVPR 2024] RCL: Reliable Continual Learning for Unified Failure Detection [paper] [code].
  Fei Zhu, Zhen Cheng, Xu-Yao Zhang, Cheng-Lin Liu, Zhaoxiang Zhang.
  

• [CVPR 2023 Highlight Paper (Top 2.5%)] OpenMix: Exploring Outlier Samples for Misclassification Detection [paper] [arxiv] [code].
  Fei Zhu, Zhen Cheng, Xu-Yao Zhang, Cheng-Lin Liu.
  

• [ECCV 2022] Rethinking Confidence Calibration for Failure Prediction [paper] [arxiv] [code].
  Fei Zhu, Zhen Cheng, Xu-Yao Zhang, Cheng-Lin Liu.
  

• [CVPR 2021 Oral Paper (Top 4%)] Prototype Augmentation and Self-Supervision for Incremental Learning [paper] [code].
  Fei Zhu, Xu-Yao Zhang, Chuang Wang, Fei Yin, Cheng-Lin Liu.
  

• [NeurIPS 2021] Class-Incremental Learning via Dual Augmentation [paper] [code].
  Fei Zhu, Zhen Cheng, Xu-Yao Zhang, Cheng-Lin Liu.
  

• [IEEE/CAA JAS 2023 Invited Reviews] Class Incremental Learning: A Review and Performance Evaluation (In Chinese) [paper].
  Fei Zhu, Xu-Yao Zhang, Cheng-Lin Liu.
  

• [TPAMI 2025] ProtoGCD: Unified and Unbiased Prototype Learning for Generalized Category Discovery [paper] [code].
  Shijie Ma, Fei Zhu, Xu-Yao Zhang, Cheng-Lin Liu.
  

• [ICLR 2025] C-CLIP: Multimodal Continual Learning for Vision-Language Model [paper].
  Wen-Zhuo Liu, Fei Zhu^📧, Longhui Wei, Qi Tian.
  

• [ICCV 2025] Federated Continual Instruction Tuning [paper].
  Haiyang Guo, Fanhu Zeng, Fei Zhu, Wenzhuo Liu, Da-Han Wang, Jian Xu, Xu-Yao Zhang, Cheng-Lin Liu.
  

• [ACL 2025] HiDe-LLaVA: Hierarchical decoupling for continual instruction tuning of multimodal large language model [paper] [code].
  Haiyang Guo, Fanhu Zeng, Ziwei Xiang, Fei Zhu, Da-Han Wang, Xu-Yao Zhang, Cheng-Lin Liu.
  

• [TNNLS 2025] Branch-Tuning: Balancing Stability and Plasticity for Continual Self-Supervised Learning [paper].
  Wenzhuo Liu, Fei Zhu, Cheng-Lin Liu.
  

• [TNNLS 2025] Average of Pruning: Improving Performance and Stability of Out-of-Distribution Detection [paper].
  Zhen Cheng, Fei Zhu, Xu-Yao Zhang, Cheng-Lin Liu.
  

• [NeurIPS 2024] MSPE: Multi-Scale Patch Embedding Prompts Vision Transformers to Any Resolution [paper].
  Wenzhuo Liu, Fei Zhu, Shijie Ma, Cheng-Lin Liu.
  

• [NeurIPS 2024] Happy: A Debiased Learning Framework for Continual Generalized Category Discovery [paper] [code].
  Shijie Ma, Fei Zhu, Zhun Zhong, Xu-Yao Zhang, Cheng-Lin Liu.
  

• [IJCV 2024] Breaking the Limits of Reliable Prediction via Generated Data [paper].
  Zhen Cheng, Fei Zhu, Xu-Yao Zhang, Cheng-Lin Liu.
  

• [PR 2024] Towards trustworthy dataset distillation [paper] [code].
  Shijie Ma, Fei Zhu, Zhen Cheng, Xu-Yao Zhang.
  

• [ECCV 2024] PILoRA: Prototype Guided Incremental LoRA for Federated Class-Incremental Learning [paper] [arxiv] [code].
  Haiyang Guo, Fei Zhu, Wenzhuo Liu, Xu-Yao Zhang, Cheng-Lin Liu.
  

• [CVPR 2024] Active Generalized Category Discovery [paper] [arxiv] [code].
  Shijie Ma, Fei Zhu, Zhun Zhong, Xu-Yao Zhang, Cheng-Lin Liu.
  

• [PR 2023] Adversarial Training with Distribution Normalization and Margin Balance [paper].
  Zhen Cheng, Fei Zhu, Xu-Yao Zhang, Cheng-Lin Liu.
  

• [Nature Communications 2022 Highlight Paper] Decoding lip language using triboelectric sensors with deep learning [paper].
  Yi-Jia Lu*, Han Tan*, Jia Cheng*, Fei Zhu, Bing Liu, Shan-Shan Wei, LinHong Ji, Zhong-Lin Wang.
  

Academic Services

• Conference Reviewer: NeurIPS, ICLR, ICML, CVPR, ICCV, ECCV

• Journal Reviewer: IEEE TIP, TNNLS, TMM, TKDE, PR, NN, IJCV

• Workshop Organizer: Trustworthy Model and Learning in Open Environment, PRCV 2024