Optimizing Data Augmentation through Bayesian Model Selection

ICLR 2026 Conference Submission19316 Authors

Keywords: Bayesian Neural Network, Variational Inference, Data Augmentation

TL;DR: We present OPTIMA, a Bayesian framework that learns augmentation and model parameters jointly, with theory and experiments showing improved accuracy, calibration, and OOD robustness over strong baselines.

Abstract:

Data Augmentation (DA) has become an essential tool to improve robustness and generalization of modern machine learning.
However, when deciding on DA strategies it is critical to choose parameters carefully, and this can be a daunting task which is traditionally left to trial-and-error or expensive optimization based on validation performance.
In this paper, we counter these limitations by proposing a novel framework for optimizing DA.
In particular, we take a probabilistic view of DA, which leads to the interpretation of augmentation parameters as model (hyper)-parameters, and the optimization of the marginal likelihood with respect to these parameters as a Bayesian model selection problem.
Due to its intractability, we derive a tractable ELBO, which allows us to optimize augmentation parameters jointly with model parameters. We provide extensive theoretical results on variational approximation quality, generalization guarantees, invariance properties, and connections to empirical Bayes.
Through experiments on computer vision tasks, we show that our approach improves calibration and yields robust performance over fixed or no augmentation.
Our work provides a rigorous foundation for optimizing \da through Bayesian principles with significant potential for robust machine learning.

Primary Area: probabilistic methods (Bayesian methods, variational inference, sampling, UQ, etc.)

Submission Number: 19316