EEC 227C: Convex Optimization and Approximation
Announcements
Description:
This course will cover the cutting-edge theory of convex optimization. Topics include gradient descent, mirror descent, multiplicative weight update, accelerated gradient descent, geometric descent, coordinate descent, smooth and nonsmooth optimization, Kelly's method and the level method, Newton's method, cubic regularization of Newton's method, accelerated cubic Newton's method, the modified Gauss-Newton method, cutting plane method, ellipsoid method, interior point method, primal-dual model of the objective function, smoothing technique, finite-sum optimization and variance reduction, optimization in relative scale, and the applications of convex optimization theory to combinatorial optimization and theorem proving.
|