A Proximal Block Coordinate Descent Algorithm for Deep Neural Network Training
Abstract
Training deep neural networks (DNNs) efficiently is a challenge due to the associated highly nonconvex optimization. The backpropagation (backprop) algorithm has long been the most widely used algorithm for gradient computation of parameters of DNNs and is used along with gradient descent-type algorithms for this optimization task. Recent work have shown the efficiency of block coordinate descent (BCD) type methods empirically for training DNNs. In view of this, we propose a novel algorithm based on the BCD method for training DNNs and provide its global convergence results built upon the powerful framework of the Kurdyka-Lojasiewicz (KL) property. Numerical experiments on standard datasets demonstrate its competitive efficiency against standard optimizers with backprop.
Cite
@article{arxiv.1803.09082,
title = {A Proximal Block Coordinate Descent Algorithm for Deep Neural Network Training},
author = {Tim Tsz-Kit Lau and Jinshan Zeng and Baoyuan Wu and Yuan Yao},
journal= {arXiv preprint arXiv:1803.09082},
year = {2018}
}
Comments
The 6th International Conference on Learning Representations (ICLR 2018), Workshop Track