On the Self-Repair Role of Astrocytes in STDP Enabled Unsupervised SNNs

TL;DR

This paper investigates how astrocytes, a type of glial cell, can enhance the self-repair and fault tolerance of unsupervised spiking neural networks trained with STDP, inspired by biological brain functions.

Contribution

It introduces a novel perspective on incorporating astrocyte-like mechanisms into SNNs to improve their fault tolerance and self-repair capabilities.

Findings

Astrocyte-inspired mechanisms enable significant self-repair in SNNs with 50-90% faults.

Fault-tolerant SNNs maintain high accuracy on MNIST and Fashion-MNIST datasets.

Biological insights improve robustness of neuromorphic computing models.

Abstract

Neuromorphic computing is emerging to be a disruptive computational paradigm that attempts to emulate various facets of the underlying structure and functionalities of the brain in the algorithm and hardware design of next-generation machine learning platforms. This work goes beyond the focus of current neuromorphic computing architectures on computational models for neuron and synapse to examine other computational units of the biological brain that might contribute to cognition and especially self-repair. We draw inspiration and insights from computational neuroscience regarding functionalities of glial cells and explore their role in the fault-tolerant capacity of Spiking Neural Networks (SNNs) trained in an unsupervised fashion using Spike-Timing Dependent Plasticity (STDP). We characterize the degree of self-repair that can be enabled in such networks with varying degree of faults ranging from 50% - 90% and evaluate our proposal on the MNIST and Fashion-MNIST datasets.