Parallel Sparse Optimization
Year: 2013
Type: dissertation
Abstract: This thesis proposes parallel and distributed algorithms for solving very largescale\nsparse optimization problems on computer clusters and clouds. Many modern\napplications problems from compressive sensing, machine learning and signal and\nimage processing involve large-scale data and can be modeled as sparse optimization\nproblems. Those problems are in such a large-scale that they can no longer\nbe processed on single workstations running single-threaded computing approaches.\nMoving to parallel/distributed/cloud computing becomes a viable option. I propose\ntwo approaches for solving these problems. The first approach is the distributed\nimplementations of a class of efficient proximal linear methods for solving convex\noptimization problems by taking advantages of the separability of the terms in the objective.\nThe second approach is a parallel greedy coordinate descent method (GRock),\nwhich greedily choose several entries to update in parallel in each iteration. I establish\nthe convergence of GRock and explain why it often performs exceptionally well for\nsparse optimization. Extensive numerical results on a computer cluster and Amazon\nEC2 demonstrate the efficiency and elasticity of my algorithms.
Author: Peng, Zhimin
Language: English
Subfield: Signal Processing
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W2296348593
