Practical Gauss-Newton Optimisation for Deep Learning
Botev, Aleksandar, Ritter, Hippolyt, Barber, David
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.
Jun-13-2017
- Country:
- Europe > United Kingdom (0.14)
- Oceania > Australia (0.14)
- Genre:
- Research Report > New Finding (0.66)
- Technology: