Regularizing Neural Networks via Adversarial Model Perturbation
arXiv:2010.04925
Abstract
Effective regularization techniques are highly desired in deep learning for alleviating overfitting and improving generalization. This work proposes a new regularization scheme, based on the understanding that the flat local minima of the empirical risk cause the model to generalize better. This scheme is referred to as adversarial model perturbation (AMP), where instead of directly minimizing the empirical risk, an alternative "AMP loss" is minimized via SGD. Specifically, the AMP loss is obtained from the empirical risk by applying the "worst" norm-bounded perturbation on each point in the parameter space. Comparing with most existing regularization schemes, AMP has strong theoretical justifications, in that minimizing the AMP loss can be shown theoretically to favour flat local minima of the empirical risk. Extensive experiments on various modern deep architectures establish AMP as a new state of the art among regularization schemes. Our code is available at https://github.com/hiyouga/AMP-Regularizer.
arXiv comment16 pages, 13 figures, accepted to CVPR2021
References in corpus (11)
- Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift
- Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks
- PyTorch: An Imperative Style, High-Performance Deep Learning Library
- Improved Regularization of Convolutional Neural Networks with Cutout
- Accurate, Large Minibatch SGD: Training ImageNet in 1 Hour
- On Calibration of Modern Neural Networks
- Reconciling modern machine learning practice and the bias-variance trade-off
- Neural Tangent Kernel: Convergence and Generalization in Neural Networks
- Three Factors Influencing Minima in SGD
- ShakeDrop Regularization for Deep Residual Learning
- Sharp Minima Can Generalize For Deep Nets