Active Predictive Coding Networks: A Neural Solution to the Problem of Learning Reference Frames and Part-Whole Hierarchies

TL;DR

Active Predictive Coding Networks (APCNs) are a novel neural architecture that learns to parse visual scenes into hierarchical parts and reference frames, combining hypernetworks and reinforcement learning for dynamic, interpretable scene understanding.

Contribution

This paper introduces APCNs, a new neural network model that learns hierarchical part-whole structures and intrinsic reference frames using hypernetworks and reinforcement learning.

Findings

APCNs successfully parse images into part-whole hierarchies.

APCNs learn compositional representations transferable to unseen classes.

APCNs enable explainable AI through dynamic parse tree generation.

Abstract

We introduce Active Predictive Coding Networks (APCNs), a new class of neural networks that solve a major problem posed by Hinton and others in the fields of artificial intelligence and brain modeling: how can neural networks learn intrinsic reference frames for objects and parse visual scenes into part-whole hierarchies by dynamically allocating nodes in a parse tree? APCNs address this problem by using a novel combination of ideas: (1) hypernetworks are used for dynamically generating recurrent neural networks that predict parts and their locations within intrinsic reference frames conditioned on higher object-level embedding vectors, and (2) reinforcement learning is used in conjunction with backpropagation for end-to-end learning of model parameters. The APCN architecture lends itself naturally to multi-level hierarchical learning and is closely related to predictive coding models of cortical function. Using the MNIST, Fashion-MNIST and Omniglot datasets, we demonstrate that APCNs can (a) learn to parse images into part-whole hierarchies, (b) learn compositional representations, and (c) transfer their knowledge to unseen classes of objects. With their ability to dynamically generate parse trees with part locations for objects, APCNs offer a new framework for explainable AI that leverages advances in deep learning while retaining interpretability and compositionality.