RESEARCH JOURNEY / CONTINUITY
How the questions
became a research direction.
This is not only a chronology of institutions. It is the progression from visual modeling to multimodal perception, efficient representation, and on-device agents.
Foundations
Artificial intelligence and visual modeling
Built foundations in AI, mathematical modeling, and research practice during undergraduate study at DLUT.
- B.Eng. in Artificial Intelligence
- National mathematical modeling first prize
First research direction
Object re-identification in real environments
Moved from general computer vision toward identity modeling, smart-city settings, and research communication.
- CVPR Smart City Challenge
- Rethinking Object ReID talk at NCCA
Multimodal perception
Identity across RGB, NIR, TIR, and tokens
Developed a systematic multimodal ReID direction through token permutation and diverse token selection.
- TOP-ReID · AAAI 2024
- Magic Tokens · CVPR 2024
Efficient representation
Language priors, expert routing, and Mamba
Expanded multimodal perception toward vision-language semantics and efficient long-sequence representation.
- IDEA · CVPR 2025
- DeMo and MambaPro · AAAI 2025
- Research internship · NUS LV Lab
PhD year 1 · Information & Communication Engineering
A connected program in multimodal intelligence
Started the first year of my PhD at the School of Information and Communication Engineering, bringing multimodal perception, efficient models, and reliable on-device agents into one research program.
- PhD Year 1 · DLUT Information & Communication Engineering
- ERA · Efficient multimodal models
- OPPO Y-Lab collaboration
- 2.5× TTFT improvement
Research vision
Reliable cloud–edge agent systems
Toward agents that coordinate multimodal perception, memory, reasoning, and action across device and cloud.
- Adaptive computation
- Long-horizon decision-making
EDUCATION & RESEARCH EXPERIENCE
Where the work happened.
Formal education, research internships, industry collaboration, presentations, and selected milestones.
Education
Ph.D. in Information and Communication Engineering
School of Information and Communication Engineering · Dalian University of Technology
First-year doctoral research connecting multimodal perception, efficient multimodal models, and reliable on-device agents.
Industry research
Research practice
OPPO Y-Lab
Collaborated on efficient multimodal GUI agents, connecting model pruning with on-device latency and package-size constraints.
Education
M.S. in Computer Science
School of Computer Science and Technology · Dalian University of Technology
Research focus: multimodal object re-identification and efficient visual representation learning.
Research
Research Intern
LV LAB · National University of Singapore
Worked under Prof. Xinchao Wang on vision and language research.
Talk
Rethinking Object ReID
National Centre for Computer Animation · Bournemouth, UK
Recommended academic presentation on the fundamentals, recent progress, and future directions of object re-identification.
Education
B.Eng. in Artificial Intelligence
School of Future Technology · Dalian University of Technology
Graduated ranked 1/94 (top 1%) with an Outstanding Undergraduate Thesis.
Achievement
CVPR 2023 Workshop Smart City Challenge
6th place
Collaborative competition result in a smart-city vision task.
Achievement
National College Student Mathematical Modeling Competition
First Prize · Team Leader
Led the team to a national first-prize result.