Singular Values for ReLU Layers

doi:10.1109/TNNLS.2019.2945113

Return

Singular Values for ReLU Layers

2020-09-01

Sören Dittmer *

Emily J. King

Peter Maaß

DOI：10.1109/TNNLS.2019.2945113

Original

Save

View PDF

Abstract

En 中文

Despite their prevalence in neural networks, we still lack a thorough theoretical characterization of rectified linear unit (ReLU) layers. This article aims to further our understanding of ReLU layers by studying how the activation function ReLU interacts with the linear component of the layer and what role this interaction plays in the success of the neural network in achieving its intended task. To this end, we introduce two new tools: ReLU singular values of operators and the Gaussian mean width of operators. By presenting, on the one hand, theoretical justifications, results, and interpretations of these two concepts and, on the other hand, numerical experiments and results of the ReLU singular values and the Gaussian mean width being applied to trained neural networks, we hope to give a comprehensive, singular-value-centric view of ReLU layers. We find that ReLU singular values and the Gaussian mean width do not only enable theoretical insights but also provide one with metrics that seem promising for practical applications. In particular, these measures can be used to distinguish correctly and incorrectly classified data as it traverses the network. We conclude by introducing two tools based on our findings: double layers and harmonic pruning.

Keywords:

Neural networks

Tools

Harmonic analysis

Learning systems

Task analysis

Measurement

Gaussian mean width

n-width

neural networks

rectified linear unit (ReLU)

singular values

AI Summary

Key information extracted from the uploaded paper, including a brief overview, abstract, background, key highlights, visual analysis, and future outlook.