Mathematics 353dl - Seminar: Advanced Topics in Discrete Applied Mathematics-Mathematics of Deep Learning

Sem:T-Deep Learning

Fall
2023
01
4.00
Luca Capogna

TU TH 9:25 AM - 10:40 AM

Smith College
MTH-353dl-01-202401
Burton 307
lcapogna@smith.edu
The course will cover topics from different parts of mathematics, with the common theme that they play some role in the design of neural networks. We will also look at some neural networks’ applications and at how mathematics is integrated. Topics will include: What is a neural network, examples and applications; Universal approximation theorems (Cybenko and others); Examples of loss functions; Gradient Descent and Stochastic Gradient descent; Generalization gap, training vs testing data; Quick review of game theory, Nash equilibrium; Generative Adversarial Networks (GAN); Unrolled GANs. Enrollment limited to 12. Juniors and seniors only. Instructor permission required.

[CE] JR/SR only

Permission is required for interchange registration during the add/drop period only.