Mathematical and Numerical Understanding of Neural Networks: From Representation to Learning Dynamics

Series
Applied and Computational Mathematics Seminar
Time
Monday, November 18, 2024 - 2:00pm for 1 hour (actually 50 minutes)
Location
Skiles 005 and https://gatech.zoom.us/j/94954654170
Speaker
Hongkai Zhao – Duke University – zhao@math.duke.eduhttps://scholars.duke.edu/person/Hongkai.Zhao
Organizer
Wei Zhu

In this talk I will present both mathematical and numerical analysis as well as experiments to study a few basic computational issues in using neural network to approximate functions: (1) the stability and accuracy, (2) the learning dynamics and computation cost, and (3) structured and balanced approximation. These issues are investigated for both approximation and optimization in asymptotic and non-asymptotic regimes.