Related papers: Explicit Regularisation in Gaussian Noise Injections

URL: http://arxiv.org/abs/2007.07368v6
Date: Tue, 19 Jan 2021 16:41:43 GMT
Title: Explicit Regularisation in Gaussian Noise Injections
Authors: Alexander Camuto, Matthew Willetts, Umut \c{S}im\c{s}ekli, Stephen Roberts, Chris Holmes
Abstract summary: We study the regularisation induced in neural networks by Gaussian noise injections (GNIs) We derive the explicit regulariser of GNIs, obtained by marginalising out the injected noise. We show analytically and empirically that such regularisation produces calibrated classifiers with large classification margins.
Score: 64.11680298737963
License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Abstract: We study the regularisation induced in neural networks by Gaussian noise injections (GNIs). Though such injections have been extensively studied when applied to data, there have been few studies on understanding the regularising effect they induce when applied to network activations. Here we derive the explicit regulariser of GNIs, obtained by marginalising out the injected noise, and show that it penalises functions with high-frequency components in the Fourier domain; particularly in layers closer to a neural network's output. We show analytically and empirically that such regularisation produces calibrated classifiers with large classification margins.

Err

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.