Automatic Data Augmentation via Invariance-Constrained Learning

TL;DR

This paper introduces an automatic data augmentation method that adapts transformations during training by formulating it as an invariance-constrained learning problem and solving it with MCMC sampling, achieving state-of-the-art results.

Contribution

It proposes a novel invariance-constrained learning framework for automatic data augmentation that dynamically controls augmentation application without prior search.

Findings

Achieves state-of-the-art results on CIFAR datasets.

Automatically adapts augmentation strategies during training.

Provides insights into underlying symmetries of learning tasks.

Abstract

Underlying data structures, such as symmetries or invariances to transformations, are often exploited to improve the solution of learning tasks. However, embedding these properties in models or learning algorithms can be challenging and computationally intensive. Data augmentation, on the other hand, induces these symmetries during training by applying multiple transformations to the input data. Despite its ubiquity, its effectiveness depends on the choices of which transformations to apply, when to do so, and how often. In fact, there is both empirical and theoretical evidence that the indiscriminate use of data augmentation can introduce biases that outweigh its benefits. This work tackles these issues by automatically adapting the data augmentation while solving the learning task. To do so, it formulates data augmentation as an invariance-constrained learning problem and leverages Monte Carlo Markov Chain (MCMC) sampling to solve it. The result is a practical algorithm that not only does away with a priori searches for augmentation distributions, but also dynamically controls if and when data augmentation is applied. Our experiments illustrate the performance of this method, which achieves state-of-the-art results in automatic data augmentation benchmarks for CIFAR datasets. Furthermore, this approach can be used to gather insights on the actual symmetries underlying a learning task.