Abstract
We consider the problem of learning a graph under the Laplacian constraint with a non-convex penalty: minimax concave penalty (MCP). For solving the MCP penalized graphical model, we design an inexact proximal difference-of-convex algorithm (DCA) and prove its convergence to critical points. We note that each subproblem of the proximal DCA enjoys the nice property that the objective function in its dual problem is continuously differentiable with a semismooth gradient. Therefore, we apply an efficient semismooth Newton method to subproblems of the proximal DCA. Numerical experiments on various synthetic and real data sets demonstrate the effectiveness of the non-convex penalty MCP in promoting sparsity. Compared with the existing state-of-the-art method, our method is demonstrated to be more efficient and reliable for learning graph Laplacian with MCP.
Original language | English |
---|---|
Pages (from-to) | 1-32 |
Number of pages | 32 |
Journal | Optimization Methods and Software |
DOIs | |
Publication status | Published - Nov 2023 |
Keywords
- difference-of-convex
- graph Laplacian estimation
- Graph learning
- non-convex penalty
- precision matrix estimation
ASJC Scopus subject areas
- Software
- Control and Optimization
- Applied Mathematics