Dynamics of Interacting Neural Networks

TL;DR

This paper analytically explores the dynamics of interacting neural networks, revealing symmetry-breaking phase transitions and demonstrating their effectiveness in adaptive competition scenarios like the Minority Game.

Contribution

It provides an analytical solution for interacting perceptrons' dynamics and shows their application to adaptive decision-making problems.

Findings

System reaches higher symmetry states with directed information flow

Symmetry-breaking phase transition occurs at higher learning rates

Interacting perceptrons trained on minority decisions develop effective strategies

Abstract

The dynamics of interacting perceptrons is solved analytically. For a directed flow of information the system runs into a state which has a higher symmetry than the topology of the model. A symmetry breaking phase transition is found with increasing learning rate. In addition it is shown that a system of interacting perceptrons which is trained on the history of its minority decisions develops a good strategy for the problem of adaptive competition known as the Bar Problem or Minority Game.