Impact of white noise in artificial neural networks trained for classification: performance and noise mitigation strategies

TL;DR

This paper investigates how Gaussian white noise affects neural network classification accuracy and evaluates noise mitigation strategies, demonstrating their effectiveness in improving performance in noisy physical neural systems.

Contribution

The study adapts and tests noise reduction techniques specifically for classification neural networks with physical noise, showing their high effectiveness.

Findings

Noise significantly impacts neural network accuracy in physical implementations.

Mitigation strategies substantially reduce the adverse effects of additive and multiplicative noise.

Effective noise mitigation enhances the reliability of physical neural network hardware.

Abstract

In recent years, the hardware implementation of neural networks, leveraging physical coupling and analog neurons has substantially increased in relevance. Such nonlinear and complex physical networks provide significant advantages in speed and energy efficiency, but are potentially susceptible to internal noise when compared to digital emulations of such networks. In this work, we consider how additive and multiplicative Gaussian white noise on the neuronal level can affect the accuracy of the network when applied for specific tasks and including a softmax function in the readout layer. We adapt several noise reduction techniques to the essential setting of classification tasks, which represent a large fraction of neural network computing. We find that these adjusted concepts are highly effective in mitigating the detrimental impact of noise.