Research Seminar by Prof Yong Luo from Wuhan University, 26 Jul 2024, CCDS Meeting Room
![Seminar_Prof Yong Luo](/images/librariesprovider118/news-events/2024/seminar_prof-yong-luo.png?sfvrsn=8f3f276d_6)
Time: 10-11:30am, July 26th Friday 2024
Venue: CCDS Meeting Room (N4-02a-35)
Title: Deep Model Fusion
Abstract:
The paradigm of deep learning has significantly evolved in recent years, moving beyond traditional supervised learning to incorporate knowledge transfer and model editing. While these emerging techniques show promise in enhancing performance, accelerated training, and reducing labeled data dependency, their full potential and scalability to large foundation models remain unexplored. This talk provides an investigation of knowledge transfer and model fusion techniques for deep neural networks, covering (1) their background, motivation, and existing approaches; (2) a taxonomy for categorizing these techniques and the formal definitions for each category; (3) our recent contributions to the field, including adaptive model ensemble, plug-and-play techniques for improved model merging, adaptive weight-based model mixing, and model merging in Pareto optimization contexts; (4) discussion of the strengths, challenges, and future directions of model fusion techniques, providing a comprehensive overview of this rapidly evolving area in deep learning.
Biography:
Yong Luo received the B.E. degree in Computer Science from the Northwestern Polytechnical University, Xi’an, China, and the D.Sc. degree in the School of Electronics Engineering and Computer Science, Peking University, Beijing, China. He was a Research Fellow with the School of Computer Science and Engineering, Nanyang Technological University, and is currently a Professor with the School of Computer Science, Wuhan University, China. His research interests are primarily on machine learning and data mining with applications to visual information understanding and analysis. He has authored or co-authored over 80 papers in top journals and prestigious conferences including Nature Communications, IEEE T-PAMI, IEEE T-NNLS, IEEE T-IP, IEEE T-KDE, IEEE T-MM, CVPR, ICCV, WWW, IJCAI and AAAI. He is serving on editorial board for IEEE T-MM. He received the IEEE Globecom 2016 Best Paper Award, and was nominated as the IJCAI 2017 Distinguished Best Paper Award. He is also a co-recipient of the IEEE TMM 2023, IEEE ICME 2019 and IEEE VCIP 2019 Best Paper Awards.