Gradient Starvation: A Learning Proclivity in Neural Networks

By Mohammad Pezeshki1,2, Sekou-Oumar Kaba1,3, Yoshua Bengio1,2, et al, 1Mila, 2 Université de Montréal, 3McGill University, 2020

In this paper, the authors examine in detail the phenomenon of gradient starvation, which was originally introduced by the same research group in 2018, for neural networks trained with the common cross-entropy loss. Gradient starvation occurs when the presence of easy-to-learn features in a dataset prevents the learning of other equally informative features, which may lead to a lack of robustness in the trained models that rely only on these few features. The authors propose a new Spectral Decoupling regularization method to combat this problem.


No more pages to load