NIPS Proceedingsβ

On the Computational Efficiency of Training Neural Networks

Part of: Advances in Neural Information Processing Systems 27 (NIPS 2014)

[PDF] [BibTeX] [Supplemental] [Reviews]

Authors

Conference Event Type: Poster

Abstract

It is well-known that neural networks are computationally hard to train. On the other hand, in practice, modern day neural networks are trained efficiently using SGD and a variety of tricks that include different activation functions (e.g. ReLU), over-specification (i.e., train networks which are larger than needed), and regularization. In this paper we revisit the computational complexity of training neural networks from a modern perspective. We provide both positive and negative results, some of them yield new provably efficient and practical algorithms for training neural networks.