Robust Large Margin Deep Neural Networks

Published

Journal Article

© 2017 IEEE. The generalization error of deep neural networks via their classification margin is studied in this paper. Our approach is based on the Jacobian matrix of a deep neural network and can be applied to networks with arbitrary nonlinearities and pooling layers, and to networks with different architectures such as feed forward networks and residual networks. Our analysis leads to the conclusion that a bounded spectral norm of the network's Jacobian matrix in the neighbourhood of the training samples is crucial for a deep neural network of arbitrary depth and width to generalize well. This is a significant improvement over the current bounds in the literature, which imply that the generalization error grows with either the width or the depth of the network. Moreover, it shows that the recently proposed batch normalization and weight normalization reparametrizations enjoy good generalization properties, and leads to a novel network regularizer based on the network's Jacobian matrix. The analysis is supported with experimental results on the MNIST, CIFAR-10, LaRED, and ImageNet datasets.

Full Text

Duke Authors

Cited Authors

  • Sokolić, J; Giryes, R; Sapiro, G; Rodrigues, MRD

Published Date

  • August 15, 2017

Published In

Volume / Issue

  • 65 / 16

Start / End Page

  • 4265 - 4280

International Standard Serial Number (ISSN)

  • 1053-587X

Digital Object Identifier (DOI)

  • 10.1109/TSP.2017.2708039

Citation Source

  • Scopus