Effect of Batch Normalization on Noise Resistant Property of Deep Learning Models

TL;DR

This paper investigates how batch normalization layers affect the noise resistance of deep learning models, revealing that batch normalization can degrade robustness against analog noise, especially with multiple layers.

Contribution

It provides a systematic analysis of the impact of batch normalization on noise resistance in deep models trained on CIFAR datasets, highlighting a negative effect.

Findings

Batch normalization reduces noise resistance in deep models.

The negative impact increases with more batch normalization layers.

Models without batch normalization are more robust to analog noise.

Abstract

The fast execution speed and energy efficiency of analog hardware has made them a strong contender for deployment of deep learning model at the edge. However, there are concerns about the presence of analog noise which causes changes to the weight of the models, leading to performance degradation of deep learning model, despite their inherent noise resistant characteristics. The effect of the popular batch normalization layer on the noise resistant ability of deep learning model is investigated in this work. This systematic study has been carried out by first training different models with and without batch normalization layer on CIFAR10 and CIFAR100 dataset. The weights of the resulting models are then injected with analog noise and the performance of the models on the test dataset is obtained and compared. The results show that the presence of batch normalization layer negatively impacts noise resistant property of deep learning model and the impact grows with the increase of the number of batch normalization layers.