The Cooperative Network Architecture: Learning Structured Networks as Representation of Sensory Patterns

TL;DR

The paper introduces the Cooperative Network Architecture (CNA), a neural model that learns structured, recurrent networks from sensory data, enhancing robustness, generalization, and the ability to encode complex patterns without supervision.

Contribution

It presents a novel architecture that dynamically assembles neural nets from learned fragments, enabling noise resilience, out-of-distribution generalization, and unsupervised learning of structured representations.

Findings

Net fragments can be learned without supervision.

CNA demonstrates robustness to noise and deformation.

The architecture enables figure completion and pattern generalization.

Abstract

We introduce the Cooperative Network Architecture (CNA), a model that represents sensory signals using structured, recurrently connected networks of neurons, termed "nets." Nets are dynamically assembled from overlapping net fragments, which are learned based on statistical regularities in sensory input. This architecture offers robustness to noise, deformation, and generalization to out-of-distribution data, addressing challenges in current vision systems from a novel perspective. We demonstrate that net fragments can be learned without supervision and flexibly recombined to encode novel patterns, enabling figure completion and resilience to noise. Our findings establish CNA as a promising paradigm for developing neural representations that integrate local feature processing with global structure formation, providing a foundation for future research on invariant object recognition.