Analysis & theory
A Discrepancy-Based Perspective on Dataset Condensation
Tong Chen, Raghavendra Selvan
arXiv 2025 · first public 2025-09-12 · arXiv 2509.10367
In one paragraph
Recasts dataset condensation as minimizing a discrepancy between the real and synthetic data distributions, showing that gradient-, distribution- and trajectory-matching objectives are instances of one framework and extending the objective beyond test accuracy to robustness and privacy criteria.
Where it sits
- Setting: Image classification
Abstract (verbatim from arXiv)
Given a dataset of finitely many elements $\mathcal{T} = \{\mathbf{x}_i\}_{i = 1}^N$, the goal of dataset condensation (DC) is to construct a synthetic dataset $\mathcal{S} = \{\tilde{\mathbf{x}}_j\}_{j = 1}^M$ which is significantly smaller ($M \ll N$) such that a model trained from scratch on $\mathcal{S}$ achieves comparable or even superior generalization performance to a model trained on $\mathcal{T}$. Recent advances in DC reveal a close connection to the problem of approximating the data distribution represented by $\mathcal{T}$ with a reduced set of points. In this work, we present a unified framework that encompasses existing DC methods and extend the task-specific notion of DC to a more general and formal definition using notions of discrepancy, which quantify the distance between probability distribution in different regimes. Our framework broadens the objective of DC beyond generalization, accommodating additional objectives such as robustness, privacy, and other desirable properties.
BibTeX (generated; prefer the venue's official entry)
@article{chen2025discrepancy,
title = {A Discrepancy-Based Perspective on Dataset Condensation},
author = {Tong Chen and Raghavendra Selvan},
journal = {arXiv preprint arXiv:2509.10367},
year = {2025}
}