English

Path-SGD: Path-Normalized Optimization in Deep Neural Networks

Machine Learning 2015-06-09 v1 Computer Vision and Pattern Recognition Neural and Evolutionary Computing Machine Learning

Abstract

We revisit the choice of SGD for training deep neural networks by reconsidering the appropriate geometry in which to optimize the weights. We argue for a geometry invariant to rescaling of weights that does not affect the output of the network, and suggest Path-SGD, which is an approximate steepest descent method with respect to a path-wise regularizer related to max-norm regularization. Path-SGD is easy and efficient to implement and leads to empirical gains over SGD and AdaGrad.

Keywords

Cite

@article{arxiv.1506.02617,
  title  = {Path-SGD: Path-Normalized Optimization in Deep Neural Networks},
  author = {Behnam Neyshabur and Ruslan Salakhutdinov and Nathan Srebro},
  journal= {arXiv preprint arXiv:1506.02617},
  year   = {2015}
}

Comments

12 pages, 5 figures

R2 v1 2026-06-22T09:49:30.461Z