Can I read On Symmetry and Initialization for Neural Networks on EtoBox?

On Symmetry and Initialization for Neural Networks by Nachum, Ido; Yehudayoff, Amir is a scholarly article available to read on EtoBox.

What is On Symmetry and Initialization for Neural Networks about?

This work provides an additional step in the theoretical understanding of neural networks. We consider neural networks with one hidden layer and show that when learning symmetric functions, one can choose initial conditions so that standard SGD training efficiently produces generalization guarantees. We empirically verify this and show that this does not hold when the initial conditions are chosen at random. The proof of convergence investigates the interaction between the two layers of the network. Our results highlight the importance of using symmetry in the design of neural networks.

Author

Nachum, Ido; Yehudayoff, Amir

Published

2019

Language

EN