Convolutional Bipartite Attractor Networks

TL;DR

This paper introduces a convolutional bipartite attractor network architecture that leverages modern deep learning techniques for image completion and super-resolution, offering an efficient alternative to generative models.

Contribution

It presents a novel convolutional bipartite attractor network with new training loss, activation function, and connectivity constraints, enabling larger and more complex problem solving.

Findings

Effective for image completion and super-resolution

Outperforms traditional attractor networks in scale

Offers a viable alternative to costly generative methods

Abstract

In human perception and cognition, a fundamental operation that brains perform is interpretation: constructing coherent neural states from noisy, incomplete, and intrinsically ambiguous evidence. The problem of interpretation is well matched to an early and often overlooked architecture, the attractor network---a recurrent neural net that performs constraint satisfaction, imputation of missing features, and clean up of noisy data via energy minimization dynamics. We revisit attractor nets in light of modern deep learning methods and propose a convolutional bipartite architecture with a novel training loss, activation function, and connectivity constraints. We tackle larger problems than have been previously explored with attractor nets and demonstrate their potential for image completion and super-resolution. We argue that this architecture is better motivated than ever-deeper feedforward models and is a viable alternative to more costly sampling-based generative methods on a range of supervised and unsupervised tasks.