Recently algorithms incorporating second order curvature information have become popular in training neural networks. The Nesterov's Accelerated Quasi-Newton (NAQ) method has shown to effectively accelerate the BFGS quasi-Newton method by incorporating the momentum term and Nesterov's accelerated gradient vector. A stochastic version of NAQ method was proposed for training of large-scale problems. However, this method incurs high stochastic variance noise. This paper proposes a stochastic variance reduced Nesterov's Accelerated Quasi-Newton method in full (SVR-NAQ) and limited (SVRLNAQ) memory forms. The performance of the proposed method is evaluated in Tensorflow on four benchmark problems - two regression and two classification problems respectively. The results show improved performance compared to conventional methods.
Cite
@article{arxiv.1910.07939,
title = {A Stochastic Variance Reduced Nesterov's Accelerated Quasi-Newton Method},
author = {Sota Yasuda and Shahrzad Mahboubi and S. Indrapriyadarsini and Hiroshi Ninomiya and Hideki Asai},
journal= {arXiv preprint arXiv:1910.07939},
year = {2020}
}