Vitality of Neural Networks under Reoccurring Catastrophic Failures

TL;DR

This paper investigates the reoccurrence of catastrophic failures in excitatory neural networks, revealing a multimodal distribution of failure times linked to neuronal plasticity and local competition, which impacts network vitality.

Contribution

It demonstrates that catastrophic failures in neural networks follow a non-Poissonian, multimodal distribution driven by neuronal plasticity and local competition mechanisms.

Findings

Failures recur on timescales of tenths to tens of seconds.

Neuronal plasticity influences failure and recovery dynamics.

Local competition among nodes affects network vitality.

Abstract

Catastrophic failures are complete and sudden collapses in the activity of large networks such as economics, electrical power grids and computer networks, which typically require a manual recovery process. Here we experimentally show that excitatory neural networks are governed by a non-Poissonian reoccurrence of catastrophic failures, where their repetition time follows a multimodal distribution characterized by a few tenths of a second and tens of seconds timescales. The mechanism underlying the termination and reappearance of network activity is quantitatively shown here to be associated with nodal time-dependent features, neuronal plasticity, where hyperactive nodes damage the response capability of their neighbors. It presents a complementary mechanism for the emergence of Poissonian catastrophic failures from damage conductivity. The effect that hyperactive nodes degenerate their neighbors represents a type of local competition which is a common feature in the dynamics of real-world complex networks, whereas their spontaneous recoveries represent a vitality which enhances reliable functionality.