Xudong Yan | Beijing Jiaotong University

About Me

👋Hi there! This is Xudong Yan (闫旭东). I am currently a Ph.D. student at Beijing Jiaotong University, supervised by Prof. Songhe Feng. Before transferring to the doctoral program, I pursued my master’s studies at BJTU. Prior to that, I received my B.Eng. degree from Qingdao University.

My research interests focus on the application of vision-language models (VLM) in the real world, including their use in zero-shot and few-shot learning, multimodal retrieval, and test-time adaptation. If you are interested in our works or would like to explore potential collaboration, please feel very free to contact me via email🙌.

Research Interests

Computer Vision: compositional zero-shot learning, few-shot learning, composed image retrieval, test-time adaptation of VLMs.

News

[Dec. 2025] Our paper about compositional zero-shot learning is submitted to CVPR 2026.
[Oct. 2025] Our paper about multimodal knowledge extraction is submitted to WWW 2026.
[Sep. 2025] Our paper about compositional zero-shot learning is accepted to NeurIPS 2025!
[Apr. 2025] Our paper about compositional zero-shot learning is accepted to IJCAI 2025!

Publications

NeurIPS

TOMCAT: Test-time Comprehensive Knowledge Accumulation for Compositional Zero-Shot Learning

Xudong Yan, Songhe Feng*.

Conference on Neural Information Processing Systems (NeurIPS), 2025.

Paper Code Early Accept

IJCAI

Leveraging MLLM Embeddings and Attribute Smoothing for Compositional Zero-Shot Learning

Xudong Yan, Songhe Feng*, Yang Zhang, Jian Yang, Yueguan Lin, Haojun Fei.

International Joint Conference on Artificial Intelligence (IJCAI), 2025.

Paper arXiv Code

Teaching

[Fall 2025] Teaching Assistant, C Programming.
[Fall 2024] Teaching Assistant, C Programming.

Honors and Awards

[Jun. 2023] Excellent Graduate of Shandong Province, Government of Shandong province.