TSynD: Targeted Synthetic Data Generation for Enhanced Medical Image Classification

TL;DR

This paper proposes a targeted synthetic data generation method for medical image classification that improves model accuracy and robustness by focusing on underrepresented data points with high epistemic uncertainty.

Contribution

It introduces a novel approach to guide generative models to produce synthetic data with high epistemic uncertainty, enhancing training effectiveness.

Findings

Improved classification accuracy on medical imaging tasks.

Enhanced robustness against data augmentations.

Greater resilience to adversarial attacks.

Abstract

The usage of medical image data for the training of large-scale machine learning approaches is particularly challenging due to its scarce availability and the costly generation of data annotations, typically requiring the engagement of medical professionals. The rapid development of generative models allows towards tackling this problem by leveraging large amounts of realistic synthetically generated data for the training process. However, randomly choosing synthetic samples, might not be an optimal strategy. In this work, we investigate the targeted generation of synthetic training data, in order to improve the accuracy and robustness of image classification. Therefore, our approach aims to guide the generative model to synthesize data with high epistemic uncertainty, since large measures of epistemic uncertainty indicate underrepresented data points in the training set. During the image generation we feed images reconstructed by an auto encoder into the classifier and compute the mutual information over the class-probability distribution as a measure for uncertainty.We alter the feature space of the autoencoder through an optimization process with the objective of maximizing the classifier uncertainty on the decoded image. By training on such data we improve the performance and robustness against test time data augmentations and adversarial attacks on several classifications tasks.