👋关于我About Me

你好，我是易慧宇。 Hello, I am Huiyu Yi.

我目前是南京大学人工智能学院博士研究生，也是 RINC 组 成员，导师为申富饶教授。我于 2025 年 6 月从南京大学人工智能学院获得人工智能学士学位，并于 2025 年 9 月开始攻读博士学位。 I am a Ph.D. student at the School of Artificial Intelligence, Nanjing University, and a member of the RINC Group under the supervision of Prof. Furao Shen. I received my B.Sc. degree in Artificial Intelligence from Nanjing University in June 2025 and started my Ph.D. in September 2025.

我喜欢研究神经网络的持续演进，并致力于实现自进化的 Agent。 I enjoy studying the continual evolution of neural networks and working toward self-evolving agents.

🌱正在寻找Currently Seeking

科研合作、实习与访问交流机会Research collaborations, internships, and visiting opportunities

我正在寻找与持续学习、类增量学习、多模态大模型和自进化 Agent 相关的科研合作、实习或访问交流机会。如果你对这些方向感兴趣，欢迎通过邮件联系我。I am currently looking for research collaborations, internships, or visiting opportunities related to continual learning, class-incremental learning, multimodal large language models, and self-evolving agents. If you are interested in these topics, please feel free to reach out by email.

huiyuyi@smail.nju.edu.cn

🔬研究兴趣Research Interests

我的研究兴趣主要集中在多模态大模型持续学习，并进一步探索能够持续学习、反思和改进的自进化 Agent。 My research focuses on continual learning for multimodal large language models, with a growing interest in self-evolving agents that can continuously learn, reflect, and improve.

Multimodal LLM Continual Learning / 多模态大模型持续学习 Self-Evolving Agents / 自进化 Agent Agentic Learning / Agent 学习 Class-Incremental Learning / 类增量学习 RSI Agent Recursive Self-Improvement Agent

📰最新动态News

  • Jun. 2026: 独立第一作者论文被 ICML 2026 接收。My first-author paper was accepted by ICML 2026.
  • Oct. 2025: 论文 MQMK 发表于 ACM MM 2025。Our MQMK paper was published at ACM MM 2025.
  • Sep. 2025: 开始在南京大学人工智能学院攻读博士学位。Started my Ph.D. at the School of Artificial Intelligence, Nanjing University.
  • Jun. 2025: 获得南京大学人工智能学院人工智能学士学位。Received my B.Sc. degree in Artificial Intelligence from Nanjing University.

📚论文成果Selected Publications

  1. Beyond Routing Saturation：从长时程类增量视角审视多模态持续指令微调中的专家路由Beyond Routing Saturation: A Long-Horizon Class-Incremental Perspective on Expert Routing in Multimodal Continual Instruction Tuning Huiyu Yi, Yongqi Xu, Bogang Zhang, Dunwei Tu, Zhiming Xu, Zhen-Hao Xie, Baile Xu, Furao Shen.
    arXiv preprint, 2026 · First author / 独立第一作者 [arXiv] [Code]
  2. Beyond Point-wise Neural Collapse：面向类增量学习的拓扑感知分层分类器Beyond Point-wise Neural Collapse: A Topology-Aware Hierarchical Classifier for Class-Incremental Learning Huiyu Yi, Zhiming Xu, Dunwei Tu, Zhicheng Wang, Baile Xu, Furao Shen.
    ICML 2026 · First author / 独立第一作者 [arXiv] [Code]
  3. Embedding Space Allocation with Angle-Norm Joint Classifiers for Few-Shot Class-Incremental LearningEmbedding Space Allocation with Angle-Norm Joint Classifiers for Few-Shot Class-Incremental Learning Dunwei Tu, Huiyu Yi, Tieyi Zhang, Ruotong Li, Furao Shen, Jian Zhao.
    Neural Networks, 2025 · Second author / 第二作者 [arXiv] [DOI]
  4. Multiple Queries with Multiple Keys：面向基于 Prompt 的持续学习的精准 Prompt 匹配范式Multiple Queries with Multiple Keys: A Precise Prompt Matching Paradigm for Prompt-based Continual Learning Dunwei Tu, Huiyu Yi, Yuchi Wang, Baile Xu, Jian Zhao, Furao Shen.
    ACM MM 2025 · Second author / 第二作者 [arXiv] [Code]
  5. Few-Shot Class-Incremental Learning with Class Centers and Contrastive Learning for Incremental Vehicle RecognitionFew-Shot Class-Incremental Learning with Class Centers and Contrastive Learning for Incremental Vehicle Recognition Huiyu Yi.
    IJCNN 2024 · First author / 独立第一作者 [DOI]

🧭教育与项目经历Education & Experience

南京大学人工智能学院 · 人工智能博士研究生Ph.D. in Artificial Intelligence · Nanjing University Sep. 2025 - Present
南京大学人工智能学院 · 人工智能学士B.Sc. in Artificial Intelligence · Nanjing University Sep. 2021 - Jun. 2025
AI 钢琴家 · 多模态音乐智能项目AI Pianist · Multimodal Music Intelligence Project 中国大学生计算机设计大赛一等奖；江苏省大学生计算机设计大赛特等奖。负责数据预处理、多任务学习代码、企业合作与项目答辩。First Prize in the China Collegiate Computer Design Competition and Grand Prize in Jiangsu. Worked on data preprocessing, multi-task learning code, industry collaboration, and project presentation.
MBTI 文本性格分类MBTI Text Personality Classification 2024 春 NLP 课程优秀展示项目。使用 BERT、多任务学习以及 NCM/NCNS 分类器完成文本分类与理论分析。Outstanding project in the Spring 2024 NLP course. Used BERT, multi-task learning, and NCM/NCNS classifiers for text classification and analysis.

💼实习经历Internship Experience

我目前正在寻找研究院、企业或实验室的实习机会，欢迎通过邮件与我联系。I am currently seeking internship opportunities at research institutes, companies, or laboratories. Please feel free to contact me by email.