Skip to main content

Generalized Alternating Direction Method of Multipliers: New Theoretical Insights and Applications

Author(s): Fang, Ethan X.; He, Bingsheng; Liu, Han; Yuan, Xiaoming

Download
To refer to this page use: http://arks.princeton.edu/ark:/88435/pr19j7s
Abstract: Recently, the alternating direction method of multipliers (ADMM) has received intensive attention from a broad spectrum of areas. The generalized ADMM (GADMM) proposed by Eckstein and Bertsekas is an efficient and simple acceleration scheme of ADMM. In this paper, we take a deeper look at the linearized version of GADMM where one of its subproblems is approximated by a linearization strategy. This linearized version is particularly efficient for a number of applications arising from different areas. Theoretically, we show the worst-case (1/𝑘) convergence rate measured by the iteration complexity (𝑘 represents the iteration counter) in both the ergodic and a nonergodic senses for the linearized version of GADMM. Numerically, we demonstrate the efficiency of this linearized version of GADMM by some rather new and core applications in statistical learning. Code packages in Matlab for these applications are also developed.
Publication Date: 6-Feb-2015
Citation: Fang, E.X., He, B., Liu, H. et al. Generalized Alternating Direction Method of Multipliers: New Theoretical Insights and Applications. Math. Prog. Comp. 7, (2015): pp. 149–187. doi:10.1007/s12532-015-0078-2
DOI: doi:10.1007/s12532-015-0078-2
ISSN: 1867-2949
EISSN: 1867-2957
Pages: 149 - 187
Type of Material: Journal Article
Journal/Proceeding Title: Mathematical Programming Computation
Version: Author's manuscript



Items in OAR@Princeton are protected by copyright, with all rights reserved, unless otherwise indicated.