Xinzi Cao

prof_pic.jpg

I am currently a Ph.D. candidate at Sun Yat-sen University and Pengcheng Laboratory, supervised by Yutong Lu. I received my Master’s degree in Computer Science from Sun Yat-sen University in 2022, and my Bachelor’s degree in Software Engineering from South China Normal University in 2020.

I’m building AI that doesn’t panic when it meets something new—teaching models to discover and recognize unknown categories in the wild. In other words, I want AI to be less like a student who only knows the test, and more like someone who can walk into a new room and still figure out what’s going on.

My research focus:

  • Open-world Learning: Developing systems that can discover, adapt, and classify unknown samples in real-world open scenarios, not just in tidy benchmark settings.

  • Generalized & Novel Category Discovery (GCD/NCD): Enabling models to learn from known categories and generalize to unseen ones—so they can recognize new things without being explicitly taught.

  • Large Models & Neural Processing: Exploring how large models can enhance learning and improve the efficiency and intelligence of neural systems.

news

Apr 14, 2026 One papers is accepted by ACL Findings 🎉.
Mar 18, 2026 Two papers are accepted by ICME 2026 😀.
Jan 12, 2026 One Technical Report on NPU Kernel generation is released in arxiv.
Dec 04, 2025 One paper on prototypes debiasing is accepted by IEEE TCSVT 🎉🎉.
Jun 26, 2025 One paper on Prompt Tunning is accepted by the journal ICCV 2025 🎉🎉.

selected publications

  1. ascendKernelGen.png
    AscendKernelGen: A Systematic Study of LLM-Based Kernel Generation for Neural Processing Units
    Xinzi Cao, Jianyang Zhai, Pengfei Li, and 17 more authors
    2026
    HTML PDF
  2. debiasgcd.png
    Margin-Aware Prototype Debiasing for Generalized Category Discovery
    Xinzi Cao, Feidiao Yang, Xiawu Zheng, and 3 more authors
    IEEE Transactions on Circuits and Systems for Video Technology, 2025
    HTML PDF
  3. allgcd.png
    AllGCD: Leveraging All Unlabeled Data for Generalized Category Discovery
    Xinzi Cao, Ke Chen, Feidiao Yang, and 3 more authors
    In Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), Oct 2025
    HTML PDF
  4. STP.png
    Efficient Event Camera Data Pretraining with Adaptive Prompt Fusion
    Quanmin Liang, Qiang Li, Shuai Liu, and 6 more authors
    In Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), Oct 2025
    HTML PDF
  5. legogcd.png
    Solving the Catastrophic Forgetting Problem in Generalized Category Discovery
    Xinzi Cao, Xiawu Zheng, Guanhong Wang, and 5 more authors
    In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2024
    DOI HTML PDF
  6. locloc.png
    LocLoc: Low-level Cues and Local-area Guides for Weakly Supervised Object Localization
    Xinzi Cao, Xiawu Zheng, Yunhang Shen, and 4 more authors
    In Proceedings of the 31st ACM International Conference on Multimedia (ACMMM), 2023
    DOI HTML PDF