Abstract:Nonsmooth composite optimization with orthogonality constraints has a broad spectrum of applications in statistical learning and data science. However, this problem is generally challenging to solve due to its non-convex and non-smooth nature. Existing solutions are limited by one or more of the following restrictions: (i) they are full gradient methods that require high computational costs in each iteration; (ii) they are not capable of solving general nonsmooth composite problems; (iii) they are infeasible methods and can only achieve the feasibility of the solution at the limit point; (iv) they lack rigorous convergence guarantees; (v) they only obtain weak optimality of critical points. In this paper, we propose \textit{\textbf{OBCD}}, a new Block Coordinate Descent method for solving general nonsmooth composite problems under Orthogonality constraints. \textit{\textbf{OBCD}} is a feasible method with low computation complexity footprints. In each iteration, our algorithm updates $k$ rows of the solution matrix ($k\geq2$ is a parameter) to preserve the constraints. Then, it solves a small-sized nonsmooth composite optimization problem under orthogonality constraints either exactly or approximately. We demonstrate that any exact block-$k$ stationary point is always an approximate block-$k$ stationary point, which is equivalent to the critical stationary point. We are particularly interested in the case where $k=2$ as the resulting subproblem reduces to a one-dimensional nonconvex problem. We propose a breakpoint searching method and a fifth-order iterative method to solve this problem efficiently and effectively. We also propose two novel greedy strategies to find a good working set to further accelerate the convergence of \textit{\textbf{OBCD}}. Finally, we have conducted extensive experiments on several tasks to demonstrate the superiority of our approach.

Inexact Coordinate Descent: Complexity and Preconditioning

A Flexible Coordinate Descent Method

Robust Block Coordinate Descent

Coordinate Descent with Arbitrary Sampling I: Algorithms and Complexity

Coordinate-Update Algorithms can Efficiently Detect Infeasible Optimization Problems

Efficiency of Coordinate Descent Methods For Structured Nonconvex Optimization

Asynchronous Coordinate Descent under More Realistic Assumptions

Iteration complexity analysis of random coordinate descent methods for $\ell_0$ regularized convex problems

Synchronous Parallel Block Coordinate Descent Method for Nonsmooth Convex Function Minimization

Randomness and Permutations in Coordinate Descent Methods

Inexact Riemannian Gradient Descent Method for Nonconvex Optimization

Block Coordinate Descent Only Converge to Minimizers

Block Coordinate Descent Methods for Structured Nonconvex Optimization with Nonseparable Constraints: Optimality Conditions and Global Convergence

Randomized Sparse Block Kaczmarz as Randomized Dual Block-Coordinate Descent

A Limited Coordination Variance-reduced Randomized Block Scheme for Composite Nonconvex Stochastic Optimization

Random block coordinate methods for inconsistent convex optimisation problems

Derivation of Coordinate Descent Algorithms from Optimal Control Theory

A Coordinate Descent Primal-Dual Algorithm and Application to Distributed Asynchronous Optimization

The Randomized Block Coordinate Descent Method in the Hölder Smooth Setting

A Block Coordinate Descent Method for Nonsmooth Composite Optimization under Orthogonality Constraints

Asynchronous Variance-Reduced Block Schemes for Composite Non-Convex Stochastic Optimization: Block-Specific Steplengths and Adapted Batch-Sizes.