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 $${\mathcal {O}}(1/k)$$O(1/k) convergence rate measured by the iteration complexity ($$k$$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.
Original language | English (US) |
---|---|
Pages (from-to) | 149-187 |
Number of pages | 39 |
Journal | Mathematical Programming Computation |
Volume | 7 |
Issue number | 2 |
DOIs | |
State | Published - Jun 18 2015 |
All Science Journal Classification (ASJC) codes
- Theoretical Computer Science
- Software
Keywords
- Alternating direction method of multipliers
- Convergence rate
- Convex optimization
- Discriminant analysis
- Statistical learning
- Variable selection