Tessellation-Filtering ReLU Neural Networks
Biggio, Battista;
2022-01-01
Abstract
We identify tessellation-filtering ReLU neural networks that, when composed with another ReLU network, keep its non-redundant tessellation unchanged or reduce it. The additional network complexity modifies the shape of the decision surface without increasing the number of linear regions. We provide a mathematical understanding of the related additional expressiveness by means of a novel measure of shape complexity by counting deviations from convexity which results in a Boolean algebraic characterization of this special class. A local representation theorem gives rise to novel approaches for pruning and decision surface analysis.File | Size | Format | |
---|---|---|---|
0463.pdf Solo gestori archivio
Description: paper online
Type: versione editoriale
Size 319.8 kB
Format Adobe PDF
|
319.8 kB | Adobe PDF | & nbsp; View / Open Request a copy |
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.