avatar

Xu YUAN(原 旭)

PhD Candidate in Computer Science, Hong Kong Polytechnic University

Biography

☀️ I’m currently a first-year PhD student at the Department of Computing, The Hong Kong Polytechnic University (PolyU), advised by Prof. Qing Li and Prof. Wenqi Fan. I got my Master’s degree from Harbin Institute of Technology, Shenzhen (HITSZ) in March 2024, under the supervision of Prof. Zheng Zhang. Before that, I received my Bachelor’s degree from Harbin Institute of Technology at Weihai in June 2021.

News

🔥 [Dec, 2024]   One paper “Instruction-guided Multi-Granularity Segmentation and Captioning with Large Multimodal Model” has been accepted by AAAI 2025. 🎉 🎉 🎉

🔥 [Mar, 2024]   One paper has been accepted by IEEE TIP! 🎉 🎉 🎉

Research Interest

🚀 [Multimodal Large Language Models]: Retrieval-Augmented Generation, Instruction Tuning

🚀 [Multimedia Learning]: Deep Hashing Retrieval, Multimodal Retrieval

🚀 [Trustworthy Maching Learning]: Adversarial Examples, Backdoor Learning

Education

  • PhD Candidate in Computer Science, 2024.09-2027.09 (expected)
    The Hong Kong Polytechnic University, advised by Prof. Qing Li and Prof. Wenqi Fan

  • M.E. in Computer Science and Technology, 2021.09 - 2024.03
    Harbin Institute of Technology, Shenzhen, advised by Prof. Zheng Zhang

  • B.E. in Computer Science and Technology, 2017.09 - 2021.06
    Harbin Institute of Technology, Weihai

Experiences

  • March. 2024 - August. 2024: Research Intern
    • TAO Technology, Alibaba Group, Beijing
    • Working on vision and language understanding

Publications

(*) beside authors’ names indicates equal contributions.

HiBench: Benchmarking LLMs Capability on Hierarchical Structure Reasoning

Zhuohang Jiang*, Pangjing Wu*, Ziran Liang*, Peter Q. Chen*, Xu Yuan*, Ye Jia*, Tu Jiancheng*, Chen Li, Peter H. F. Ng, Li Qing

preprint, 2025.

Instruction-guided Multi-Granularity Segmentation and Captioning with Large Multimodal Model

Xu Yuan*, Li Zhou*, Zenghui Sun, Zikun Zhou, and Jinsong Lan

The 39th Association for the Advancement of Artificial Intelligence (AAAI), 2025. [CCF-A]

This work is done during an internship at Alibaba Group.

BadCM: Invisible Backdoor against Cross-Modal Learning

Zheng Zhang*, Xu Yuan*, Lei Zhu, Jingkuan Song, and Liqiang Nie

IEEE Transactions on Image Processing (TIP), 33: 2558-2571, 2024. [CCF-A]

Semantic-Aware Adversarial Training for Reliable Deep Hashing

Xu Yuan, Zheng Zhang, Xunguang Wang, and Lin Wu

IEEE Transactions on Information Forensics and Security (TIFS), 18: 4681-4694, 2023. [CCF-A]

Academic Service

  • Serving as a conference reviewer of ACMMM 2025, KDD 2025 and ICDE 2025.
  • Serving as a journal reviewer of TMM, TIFS, TKDD, and Information Fusion.

Honors and Awards

  • PolyU Presidential PhD Fellowship, 2024
  • Outstanding Master Thesis Award, Harbin Institute of Technology, 2024
  • Binxing Fang Scholarship, 2023
  • Outstanding Graduate, Harbin Institute of Technology, 2021.
  • Outstanding Bachelor Thesis Award, Harbin Institute of Technology, 2021.
  • First price, the 19th National Undergraduate Robotics Contest (ROBOCON), China, 2020.