Toward a Mathematical Theory of Deep Learning: Lessons from Personal Research

Abstract: A century ago, breakthroughs like relativity and quantum mechanics emerged from or developed alongside rigorous mathematical theories. Today's AI revolution presents a stark contrast: progress remains predominantly empirical while mathematical theory lags significantly behind. In this talk, I will share perspectives on current efforts to establish theoretical foundations for deep lear...

报告人：Weijie Su 苏炜杰 University of Pennsylvania

Bound states in the continuum:An introduction

Abstract: Bound states in the continuum (BICs) are wave modes that remain perfectly confined even though their eigen-frequencies lie inside the radiation continuum. After a brief review of classical wave confinement, I will introduce the BIC concept and explain how it defies the usual dichotomy between discrete, non-radiating spectra and continuous, leaky spectra.Throughout the talk I will conn...

报告人：吴畏 (吉林大学数学学院)

Geometric Structures of Transmission Eigenfunctions and Super-resolution Imaging

Abstract: This talk is concerned with the inverse acoustic and electromagnetic scattering problem. We use the interior resonant modes to develop an inverse scattering scheme for imaging the scatterer.Inviter: 李宏

报告人：何酉子 (深圳北理莫斯科大学)

From Industrial Software to Artificial Intelligence: Intelligent Engines of Scientific Breakthroughs and Technological Innovations

Abstract: Industrial software is the weakest component in the entire Chinese industrial system. Its research and application in China lag far behind those in developed Western countries, with over 80% of key industrial software reliant on imports. It is currently a critical technology that has challenged scientific breakthroughs and technological innovations in this country, particularly with t...

报告人：Zhangxing Chen

Instance-dependent Sample Complexity for Bilinear Saddle Point Optimization with Noisy Feedback: An LP-Based Approach

Abstract: In this work, we study the sample complexity of finding a Nash Equilibrium (NE) of two-player zero-sum matrix games with noisy feedback. We propose a novel algorithm based on resolving linear programs (LP), achieving an $\veps$-approximate NE with sample complexity $O(\min\{\frac{m_1m_2}{\epsilon^2}\log\frac{m_1m_2}{\epsilon}, \frac{m_1m_2}{\epsilon\delta^2}\log(\frac{m_1m_2}{\epsilon...

报告人：Jiashuo Jiang(Hong Kong University of Science and Technology)