A simple intelligent adaptive network

TL;DR

This paper introduces a thermodynamics-inspired self-adaptive network that dynamically reconfigures itself using macroscopic environmental feedback to maintain desired topological landscapes, demonstrating adaptability in complex, changing environments.

Contribution

It presents a novel self-adaptive network model based on thermodynamics principles that adapts using environmental entropy estimates, without relying on data-driven learning architectures.

Findings

Achieves adaptation in various constrained scenarios.

Displays a unique power law distinguishing it from memoryless systems.

Potential applications in brain and communication networks.

Abstract

For real-world complex system constantly enduring perturbation, to achieve survival goal in changing yet unknown environments, the central problem is constantly adapting themself to external environments according to environmental feedback. Such adaptability is considered the nature of general intelligence. Inspired by thermodynamics, we develop a self-adaptive network utilizing only macroscopic information to achieve desired landscape through reconfiguring itself in unknown environments. By continuously estimating environment entropy, our network can adaptively realize desired landscape represented by topological measures. Our network achieves adaptation under several scenarios, including confinement on phase space and geographic constraint. A unique power law distinguishes our network from memoryless systems. Furthermore, our simple strategy could enable brain network and communication network to adaptively maintain essential topological characteristics. Compared to data-driven methods, our self-adaptive network is understandable without careful choice of learning architectures and parameters. Our self-adaptive network could help to understand adaptive intelligence through the lens of thermodynamics.