Learning in Wilson-Cowan model for metapopulation

TL;DR

This paper introduces a novel learning algorithm based on the Wilson-Cowan metapopulation model, integrating neural mass dynamics with stable attractors to achieve high accuracy in image and text classification tasks.

Contribution

It presents a new approach that combines Wilson-Cowan neural mass models with stable attractors to enable effective learning in neural networks.

Findings

Achieves high classification accuracy on MNIST, Fashion MNIST, CIFAR-10, TF-FLOWERS, and IMDB datasets.

Demonstrates that minimal modifications to the Wilson-Cowan model can produce novel dynamics.

Integrates neural mass models with modern architectures like CNNs and transformers.

Abstract

The Wilson-Cowan model for metapopulation, a Neural Mass Network Model, treats different subcortical regions of the brain as connected nodes, with connections representing various types of structural, functional, or effective neuronal connectivity between these regions. Each region comprises interacting populations of excitatory and inhibitory cells, consistent with the standard Wilson-Cowan model. By incorporating stable attractors into such a metapopulation model's dynamics, we transform it into a learning algorithm capable of achieving high image and text classification accuracy. We test it on MNIST and Fashion MNIST, in combination with convolutional neural networks, on CIFAR-10 and TF-FLOWERS, and, in combination with a transformer architecture (BERT), on IMDB, always showing high classification accuracy. These numerical evaluations illustrate that minimal modifications to the Wilson-Cowan model for metapopulation can reveal unique and previously unobserved dynamics.