Correlated patterns in non-monotonic graded-response perceptrons

TL;DR

This paper investigates the capacity of graded-response perceptrons with non-monotonic input-output relations, emphasizing that output pattern structure primarily influences performance in storing biased, correlated patterns.

Contribution

It introduces an analysis of how output pattern structure affects the capacity of non-monotonic graded-response perceptrons with biased, correlated data.

Findings

Output pattern structure is crucial for perceptron performance.

Optimal capacity depends mainly on output pattern structure.

Non-monotonic relations influence storage capacity significantly.

Abstract

The optimal capacity of graded-response perceptrons storing biased and spatially correlated patterns with non-monotonic input-output relations is studied. It is shown that only the structure of the output patterns is important for the overall performance of the perceptrons.