# Learning robust visual representations using data augmentation invariance

## TL;DR

This paper explores how data augmentation invariance improves the robustness of deep neural network representations to identity-preserving transformations, aligning artificial models more closely with biological visual invariance.

## Contribution

It introduces data augmentation invariance, an unsupervised learning objective that enhances invariance in neural network representations with minimal additional training time.

## Key findings

- Increases invariance to image transformations by 10%
- Maintains similar categorization accuracy
- Provides a simple and efficient training enhancement

## Abstract

Deep convolutional neural networks trained for image object categorization have shown remarkable similarities with representations found across the primate ventral visual stream. Yet, artificial and biological networks still exhibit important differences. Here we investigate one such property: increasing invariance to identity-preserving image transformations found along the ventral stream. Despite theoretical evidence that invariance should emerge naturally from the optimization process, we present empirical evidence that the activations of convolutional neural networks trained for object categorization are not robust to identity-preserving image transformations commonly used in data augmentation. As a solution, we propose data augmentation invariance, an unsupervised learning objective which improves the robustness of the learned representations by promoting the similarity between the activations of augmented image samples. Our results show that this approach is a simple, yet effective and efficient (10 % increase in training time) way of increasing the invariance of the models while obtaining similar categorization performance.