Mirror Descent for Large-Scale Convex Deterministic and Stochastic Optimization
Séminaire Géométrie-Images: Calcul des variations
5/12/2013 - 15:30 Anatoli Iouditski (Université Grenoble I)
1. Proximal Setup 2. Basic Mirror Descent 3. Favorable Geometry Case 4. Stochastic Case 5. Utilizing Problem's Structure: Mirror Prox 6. Application: O(1/t) Nonsmooth Convex Minimization, 7. Acceleration by Randomization