DocumentCode :
1630207
Title :
Distributed strongly convex optimization
Author :
Tsianos, Konstantinos I. ; Rabbat, Michael G.
Author_Institution :
Dept. of Electr. & Comput. Eng., McGill Univ., Montreal, QC, Canada
fYear :
2012
Firstpage :
593
Lastpage :
600
Abstract :
A lot of effort has been invested into characterizing the convergence rates of gradient based algorithms for non-linear convex optimization. Recently, motivated by large datasets and problems in machine learning, the interest has shifted towards distributed optimization. In this work we present a distributed algorithm for strongly convex constrained optimization. Each node in a network of n computers converges to the optimum of a strongly convex, L-Lipchitz continuous, separable objective at a rate O(log (√n T)/T) where T is the number of iterations. This rate is achieved in the online setting where the data is revealed one at a time to the nodes, and in the batch setting where each node has access to its full local dataset from the start. The same convergence rate is achieved in expectation when the subgradients used at each node are corrupted with additive zero-mean noise.
Keywords :
distributed algorithms; gradient methods; optimisation; L-Lipchitz continuous objective; additive zero-mean noise; convergence rate; convex constrained optimization; distributed algorithm; distributed optimization; gradient based algorithm; machine learning; nonlinear convex optimization; Convergence; Convex functions; Cost function; Loss measurement; Program processors; Projection algorithms;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Communication, Control, and Computing (Allerton), 2012 50th Annual Allerton Conference on
Conference_Location :
Monticello, IL
Print_ISBN :
978-1-4673-4537-8
Type :
conf
DOI :
10.1109/Allerton.2012.6483272
Filename :
6483272
Link To Document :
بازگشت