Do deep nets really need weight decay and dropout?

Abstract

The impressive success of modern deep neural networks on computer vision tasks has been achieved through models of very large capacity compared to the number of available training examples. This overparameterization is often said to be controlled with the help of different regularization techniques, mainly weight decay and dropout. However, since these techniques reduce the effective capacity of the model, typically even deeper and wider architectures are required to compensate for the reduced capacity. Therefore, there seems to be a waste of capacity in this practice. In this paper we build upon recent research that suggests that explicit regularization may not be as important as widely believed and carry out an ablation study that concludes that weight decay and dropout may not be necessary for object recognition if enough data augmentation is introduced.

arXiv commentMinor changes: more explicit legend of the figure, clearer second paragraph of the introduction, details of weight decay and dropout in the appendix

References in corpus (2)

• Very Deep Convolutional Networks for Large-Scale Image Recognition
• Wide Residual Networks

Cited by in corpus (5)

• A disciplined approach to neural network hyper-parameters: Part 1 -- learning rate, batch size, momentum, and weight decay
• Further advantages of data augmentation on convolutional neural networks
• Data augmentation instead of explicit regularization
• Uses and Abuses of the Cross-Entropy Loss: Case Studies in Modern Deep Learning
• Adversarial Learning of General Transformations for Data Augmentation