Parallel dynamics of the asymmetric extremely diluted Ashkin-Teller neural network

TL;DR

This paper investigates the parallel dynamics of an asymmetric extremely diluted Ashkin-Teller neural network, deriving evolution equations and analyzing how four-spin couplings and temperature affect its retrieval capabilities.

Contribution

It introduces a detailed analysis of the network's dynamics, including derivation of evolution equations at different temperatures and insights into the impact of four-spin couplings.

Findings

Four-spin coupling enhances retrieval quality.

Derived evolution equations for order parameters.

Analyzed effects of temperature on network dynamics.

Abstract

The parallel dynamics of the asymmetric extremely diluted Ashkin-Teller neural network is studied using signal-to-noise analysis techniques. Evolution equations for the order parameters are derived, both at zero and finite temperature. The retrieval properties of the network are discussed in terms of the four-spin coupling strength and the temperature. It is shown that the presence of a four-spin coupling enhances the retrieval quality.