2019/10/30 by Ali Kavis, Kavis, Ali, Kfir Y. Levy +5 · 6 citations
Computer Science · Decision Sciences · Engineering · Mathematics · #Advanced Bandit Algorithms Research #Advanced Optimization Algorithms Research #FOS: Computer and information sciences #FOS: Mathematics #Machine Learning (cs.LG) #Optimization and Control (math.OC) #Sparse and Compressive Sensing Techniques #Stochastic Gradient Optimization Techniques
paper · pdf · doi:10.48550/arxiv.1910.13857
openalex publication_date 2019/10/30 · openalex created_date 2019/11/08 · openalex updated_date 2026/07/28
We propose a novel adaptive, accelerated algorithm for the stochastic\nconstrained convex optimization setting. Our method, which is inspired by the\nMirror-Prox method, \simultaneously achieves the optimal rates for\nsmooth/non-smooth problems with either deterministic/stochastic first-order\noracles. This is done without any prior knowledge of the smoothness nor the\nnoise properties of the problem. To the best of our knowledge, this is the\nfirst adaptive, unified algorithm that achieves the optimal rates in the\nconstrained setting. We demonstrate the practical performance of our framework\nthrough extensive numerical experiments.\n