Thermodynamic Neural Network

TL;DR

This paper introduces a thermodynamically inspired neural network model that self-organizes through local interactions, balancing reversible and irreversible processes to emulate multiscale thermodynamic evolution in open systems.

Contribution

It presents a novel neural network framework integrating thermodynamic principles with local interactions, enabling self-organization and multiscale dynamics in both isolated and driven conditions.

Findings

Networks exhibit multiscale dynamics and self-organization.

Externally driven networks efficiently connect external potentials.

The model demonstrates thermodynamic evolution of organization.

Abstract

A thermodynamically motivated neural network model is described that self-organizes to transport charge associated with internal and external potentials while in contact with a thermal reservoir. The model integrates techniques for rapid, large-scale, reversible, conservative equilibration of node states and slow, small-scale, irreversible, dissipative adaptation of the edge states as a means to create multiscale order. All interactions in the network are local and the network structures can be generic and recurrent. Isolated networks show multiscale dynamics, and externally driven networks evolve to efficiently connect external positive and negative potentials. The model integrates concepts of conservation, potentiation, fluctuation, dissipation, adaptation, equilibration and causation to illustrate the thermodynamic evolution of organization in open systems. A key conclusion of the work is that the transport and dissipation of conserved physical quantities drives the self-organization of open thermodynamic systems.