Convolutional Neural Networks with Dynamic Regularization

TL;DR

This paper introduces a dynamic regularization technique for CNNs that adjusts regularization strength based on training loss, improving generalization without manual tuning.

Contribution

It presents a novel adaptive regularization method that automatically balances overfitting and underfitting during CNN training.

Findings

Outperforms existing regularization methods on standard architectures

Automatically adjusts regularization strength based on training loss

Enhances model generalization capabilities

Abstract

Regularization is commonly used for alleviating overfitting in machine learning. For convolutional neural networks (CNNs), regularization methods, such as DropBlock and Shake-Shake, have illustrated the improvement in the generalization performance. However, these methods lack a self-adaptive ability throughout training. That is, the regularization strength is fixed to a predefined schedule, and manual adjustments are required to adapt to various network architectures. In this paper, we propose a dynamic regularization method for CNNs. Specifically, we model the regularization strength as a function of the training loss. According to the change of the training loss, our method can dynamically adjust the regularization strength in the training procedure, thereby balancing the underfitting and overfitting of CNNs. With dynamic regularization, a large-scale model is automatically regularized by the strong perturbation, and vice versa. Experimental results show that the proposed method can improve the generalization capability on off-the-shelf network architectures and outperform state-of-the-art regularization methods.