English

AutoClip: Adaptive Gradient Clipping for Source Separation Networks

Audio and Speech Processing 2020-07-30 v1 Machine Learning Sound Machine Learning

Abstract

Clipping the gradient is a known approach to improving gradient descent, but requires hand selection of a clipping threshold hyperparameter. We present AutoClip, a simple method for automatically and adaptively choosing a gradient clipping threshold, based on the history of gradient norms observed during training. Experimental results show that applying AutoClip results in improved generalization performance for audio source separation networks. Observation of the training dynamics of a separation network trained with and without AutoClip show that AutoClip guides optimization into smoother parts of the loss landscape. AutoClip is very simple to implement and can be integrated readily into a variety of applications across multiple domains.

Keywords

Cite

@article{arxiv.2007.14469,
  title  = {AutoClip: Adaptive Gradient Clipping for Source Separation Networks},
  author = {Prem Seetharaman and Gordon Wichern and Bryan Pardo and Jonathan Le Roux},
  journal= {arXiv preprint arXiv:2007.14469},
  year   = {2020}
}

Comments

Accepted at 2020 IEEE International Workshop on Machine Learning for Signal Processing, Sept.\ 21--24, 2020, Espoo, Finland

R2 v1 2026-06-23T17:28:38.088Z