Norm-Based Capacity Control in Neural Networks

TL;DR

This paper explores how norm constraints affect the capacity and convexity of feed-forward neural networks, providing insights into their theoretical properties.

Contribution

It offers a new analysis of the capacity and convexity of norm-constrained neural networks, expanding understanding of their theoretical behavior.

Findings

Norm constraints influence network capacity

Convexity properties depend on norm choices

Characterization of norm-constrained networks provided

Abstract

We investigate the capacity, convexity and characterization of a general family of norm-constrained feed-forward networks.