Saved in:
Bibliographic Details
Main Authors: Jin, Ruhui, Mixon, Dustin G., Villar, Soledad
Format: Preprint
Published: 2018
Subjects:
Online Access:https://arxiv.org/abs/1803.09319
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866917282996813824
author Jin, Ruhui
Mixon, Dustin G.
Villar, Soledad
author_facet Jin, Ruhui
Mixon, Dustin G.
Villar, Soledad
contents Deep neural networks are often used to implement powerful generative models for real-world data. Notable applications include image denoising, as well as other classical inverse problems like compressed sensing and super-resolution. To provide a rigorous but simplified analysis of generative models, in this work, we introduce an elegant theoretical framework based on spherical harmonics, namely \textbf{SUNLayer}. Our theoretical framework identifies explicit conditions on activation functions that guarantee denoising under local optimization. Numerical experiments examine the theoretical properties on commonly used activation functions and demonstrate their stable denoising performance.
format Preprint
id arxiv_https___arxiv_org_abs_1803_09319
institution arXiv
publishDate 2018
record_format arxiv
spellingShingle SUNLayer: Stable denoising with generative networks
Jin, Ruhui
Mixon, Dustin G.
Villar, Soledad
Machine Learning
Deep neural networks are often used to implement powerful generative models for real-world data. Notable applications include image denoising, as well as other classical inverse problems like compressed sensing and super-resolution. To provide a rigorous but simplified analysis of generative models, in this work, we introduce an elegant theoretical framework based on spherical harmonics, namely \textbf{SUNLayer}. Our theoretical framework identifies explicit conditions on activation functions that guarantee denoising under local optimization. Numerical experiments examine the theoretical properties on commonly used activation functions and demonstrate their stable denoising performance.
title SUNLayer: Stable denoising with generative networks
topic Machine Learning
url https://arxiv.org/abs/1803.09319