Introducing topography in convolutional neural networks

TL;DR

This paper introduces a topographic inductive bias in CNNs inspired by brain organization, improving memory efficiency and robustness without sacrificing performance across multiple datasets and models.

Contribution

A novel topographic loss and implementation for CNNs, demonstrating improved resistance to pruning and generalizability across tasks and architectures.

Findings

Equivalent performance to benchmarks on vision and audio tasks

Enhanced resistance to pruning in CNNs

Applicable to various topographic organizations

Abstract

Parts of the brain that carry sensory tasks are organized topographically: nearby neurons are responsive to the same properties of input signals. Thus, in this work, inspired by the neuroscience literature, we proposed a new topographic inductive bias in Convolutional Neural Networks (CNNs). To achieve this, we introduced a new topographic loss and an efficient implementation to topographically organize each convolutional layer of any CNN. We benchmarked our new method on 4 datasets and 3 models in vision and audio tasks and showed equivalent performance to all benchmarks. Besides, we also showcased the generalizability of our topographic loss with how it can be used with different topographic organizations in CNNs. Finally, we demonstrated that adding the topographic inductive bias made CNNs more resistant to pruning. Our approach provides a new avenue to obtain models that are more memory efficient while maintaining better accuracy.