Predictive Coding Graphs are a Superset of Feedforward Neural Networks

TL;DR

This paper demonstrates that predictive coding graphs (PCGs) encompass feedforward neural networks, positioning PCGs as a broader framework that enhances understanding of neural network topology in machine learning.

Contribution

It proves mathematically that PCGs are a superset of feedforward neural networks, linking neuroscience-inspired models with modern ML architectures.

Findings

PCGs mathematically include feedforward neural networks.

Highlights the importance of network topology in ML.

Supports non-hierarchical neural network research.

Abstract

Predictive coding graphs (PCGs) are a recently introduced generalization to predictive coding networks, a neuroscience-inspired probabilistic latent variable model. Here, we prove how PCGs define a mathematical superset of feedforward artificial neural networks (multilayer perceptrons). This positions PCNs more strongly within contemporary machine learning (ML), and reinforces earlier proposals to study the use of non-hierarchical neural networks for ML tasks, and more generally the notion of topology in neural networks.