CS 5728

Optimization for AI

Course Description

Course information provided by the 2026-2027 Catalog.

From fitting regression models to training neural networks, many problems in artificial intelligence can be cast as optimization problems. This course studies the fundamentals of optimization as they apply to artificial intelligence. Much of the focus will be on convex optimization, covering topics such as convex analysis, gradient descent for constrained and unconstrained problems, proximal gradient descent, variants of Newton’s method, and the Frank-Wolfe algorithm. The treatment will be at a fairly sophisticated level, giving sound justifications for the algorithms studied.

Prerequisites background equivalent to: MATH 1910 and MATH 2940.

Enrollment Information Enrollment limitations vary by term.

Last 4 Terms Offered 2026FA, 2026SP

Learning Outcomes

• Formulate and analyze unconstrained and constrained nonlinear optimization problems, identifying convexity, feasibility, and optimality conditions.
• Apply and implement first- and second-order optimization algorithms, and evaluate their convergence behavior in practice.
• Model and solve real-world applications using nonlinear optimization techniques in areas such as machine learning, engineering, and operations research.

View Enrollment Information