Deterministic and stochastic first order algorithms of large-scale convex optimization
Résumé
Deterministic and stochastic first order algorithms of large-scale convex optimization Syllabus: First Order Methods of proximal type Nonsmooth Black Box setting: deterministic and stochastic Mirror Descent algorithm (MD). Convex-Concave Saddle Point problems via MD. Utilizing problem's structure: Mirror Prox algorithm (MP). Smooth/Bilinear Saddle Point reformulations of convex problems: calculus and examples. The Mirror Prox algorithm. Favorable geometry domains and good proximal setups. Conditional Gradient type First Order Methods for problems with difficult geometry: Convex problems with difficult geometry. Smooth minimization, norm-regularized smooth minimization. Nonsmooth minimization.