Generating High Fidelity Synthetic Data via Coreset selection and Entropic Regularization

TL;DR

This paper introduces a method combining coreset selection and entropic regularization to generate high-fidelity synthetic data, improving data quality and model accuracy in semi-supervised learning.

Contribution

It proposes a novel approach using coresets and entropic regularization with energy-based models to select high-quality synthetic samples for enhanced learning.

Findings

Selected samples improve model accuracy more than using all synthetic data

The method effectively identifies high-fidelity samples from generative models

Augmenting datasets with selected samples boosts semi-supervised learning performance

Abstract

Generative models have the ability to synthesize data points drawn from the data distribution, however, not all generated samples are high quality. In this paper, we propose using a combination of coresets selection methods and ``entropic regularization'' to select the highest fidelity samples. We leverage an Energy-Based Model which resembles a variational auto-encoder with an inference and generator model for which the latent prior is complexified by an energy-based model. In a semi-supervised learning scenario, we show that augmenting the labeled data-set, by adding our selected subset of samples, leads to better accuracy improvement rather than using all the synthetic samples.