Log In Sign Up

How Does Sharpness-Aware Minimization Minimize Sharpness?

by   Kaiyue Wen, et al.

Sharpness-Aware Minimization (SAM) is a highly effective regularization technique for improving the generalization of deep neural networks for various settings. However, the underlying working of SAM remains elusive because of various intriguing approximations in the theoretical characterizations. SAM intends to penalize a notion of sharpness of the model but implements a computationally efficient variant; moreover, a third notion of sharpness was used for proving generalization guarantees. The subtle differences in these notions of sharpness can indeed lead to significantly different empirical results. This paper rigorously nails down the exact sharpness notion that SAM regularizes and clarifies the underlying mechanism. We also show that the two steps of approximations in the original motivation of SAM individually lead to inaccurate local conclusions, but their combination accidentally reveals the correct effect, when full-batch gradients are applied. Furthermore, we also prove that the stochastic version of SAM in fact regularizes the third notion of sharpness mentioned above, which is most likely to be the preferred notion for practical performance. The key mechanism behind this intriguing phenomenon is the alignment between the gradient and the top eigenvector of Hessian when SAM is applied.


page 1

page 2

page 3

page 4


Input Hessian Regularization of Neural Networks

Regularizing the input gradient has shown to be effective in promoting t...

Towards Understanding Sharpness-Aware Minimization

Sharpness-Aware Minimization (SAM) is a recent training method that reli...

Model Generalization: A Sharpness Aware Optimization Perspective

Sharpness-Aware Minimization (SAM) and adaptive sharpness-aware minimiza...

Sharpness-Aware Minimization for Efficiently Improving Generalization

In today's heavily overparameterized models, the value of the training l...

signADAM: Learning Confidences for Deep Neural Networks

In this paper, we propose a new first-order gradient-based algorithm to ...

SADT: Combining Sharpness-Aware Minimization with Self-Distillation for Improved Model Generalization

Methods for improving deep neural network training times and model gener...

An SDE for Modeling SAM: Theory and Insights

We study the SAM (Sharpness-Aware Minimization) optimizer which has rece...