English

Linearly-constrained nonsmooth optimization for training autoencoders

Optimization and Control 2022-04-22 v3

Abstract

A regularized minimization model with l1l_1-norm penalty (RP) is introduced for training the autoencoders that belong to a class of two-layer neural networks. We show that the RP can act as an exact penalty model which shares the same global minimizers, local minimizers, and d(irectional)-stationary points with the original regularized model under mild conditions. We construct a bounded box region that contains at least one global minimizer of the RP, and propose a linearly constrained regularized minimization model with l1l_1-norm penalty (LRP) for training autoencoders. A smoothing proximal gradient algorithm is designed to solve the LRP. Convergence of the algorithm to a generalized d-stationary point of the RP and LRP is delivered. Comprehensive numerical experiments convincingly illustrate the efficiency as well as the robustness of the proposed algorithm.

Keywords

Cite

@article{arxiv.2103.16232,
  title  = {Linearly-constrained nonsmooth optimization for training autoencoders},
  author = {Wei Liu and Xin Liu and Xiaojun Chen},
  journal= {arXiv preprint arXiv:2103.16232},
  year   = {2022}
}

Comments

to be published in "SIAM Journal on Optimization"

R2 v1 2026-06-24T00:41:11.760Z