Practical Gauss-Newton Optimisation for Deep Learning
Machine Learning
2017-06-14 v2
Abstract
We present an efficient block-diagonal ap- proximation to the Gauss-Newton matrix for feedforward neural networks. Our result- ing algorithm is competitive against state- of-the-art first order optimisation methods, with sometimes significant improvement in optimisation performance. Unlike first-order methods, for which hyperparameter tuning of the optimisation parameters is often a labo- rious process, our approach can provide good performance even when used with default set- tings. A side result of our work is that for piecewise linear transfer functions, the net- work objective function can have no differ- entiable local maxima, which may partially explain why such transfer functions facilitate effective optimisation.
Cite
@article{arxiv.1706.03662,
title = {Practical Gauss-Newton Optimisation for Deep Learning},
author = {Aleksandar Botev and Hippolyt Ritter and David Barber},
journal= {arXiv preprint arXiv:1706.03662},
year = {2017}
}
Comments
ICML 2017