MS12
Title: Understanding the Learning of Deep Networks: Expressivity, Optimization, and Generalization
Session 1
| Name of the speaker |
Title |
University |
| Fenglei Fan |
Width and Depth Equivalence: from the Perspective of Dynamic System |
Department of Data Science, City University of Hong Kong |
| Juncai He |
On the Expressivity of Neural Networks and Its Applications |
Yau Mathematical Sciences Center, Tsinghua University |
| Wei Huang |
On the Comparison between Multi-modal and Single-modal Contrastive Learning |
RIKEN Center of Advanced Intelligence, Japan |
Session 2
| Name of the speaker |
Title |
University |
| Enming Liang |
Homeomorphic Projection to Ensure Neural-Network Solution Feasibility for Constrained Optimization |
Department of Data Science, City University of Hong Kong |
| Fanghui Liu |
Norm-based capacity in machine learning: Generalization, deterministic equivalence and function spaces |
Department of Mathematics, University of Warwick |
| Atsushi Nitanda |
Propagation of Chaos for Mean-Field Langevin Dynamics and its Application to Model Ensemble |
A*STAR Centre for Frontier AI and Research (A*STAR CFAR) |
Session 3
| Name of the speaker |
Title |
University |
| Lei Shi |
Learning Theory of Classification with Deep Neural Networks |
School of Mathematical Sciences, Fudan University |
| Jiaye Teng |
A Mathematically Provable Two-Stage Training Dynamics in Transformers |
Shanghai University of Finance and Economics |
| Jinming Weng |
Randomized Orthogonal Matching Pursuit Algorithm with Adaptive Partial Selection for Sparse Signal Recovery |
Jilin University |
Session 4
| Name of the speaker |
Title |
University |
| Xiaotong Yuan |
Derandomized Online-to-Non-convex Conversion for Stochastic Weakly Convex Optimization |
School of Intelligence Science and Technology |
| Shao-Qun Zhang |
Recent Studies on Spiking Neural Networks |
Nanjing University |
| Shijun Zhang |
Tackling High-Frequency Challenges: From Shallow to Multi-Layer Neural Networks |
Department of Applied Mathematics, Hong Kong Polytechnic University |
| Chenghao Liu |
Characterizing ResNet’s Universal Approximation Capability |
Department of Data Science, City University of Hong Kong |
|
