A central problem in machine learning is to learn from data (``big''...
SE/EC/ME 724 Advanced Optimization Theory and Methods
SE/EC 524/674 or consent of instructor.
Complements SE/EC 524/674 by introducing advanced optimization techniques. Emphasis on nonlinear optimization and recent developments in the field. Topics include: unconstrained optimization methods such as gradient and incremental gradient, conjugate direction, Newton and quasi-Newton methods; constrained optimization methods such as projection, feasible directions, barrier and interior point methods; duality theory and methods; convex duality; and stochastic approximation algorithms. Introduction to modern convex optimization including semi-definite programming, conic programming, and robust optimization. Applications drawn from control, production and capacity planning, resource allocation, communication and sensor networks, and bioinformatics.