DANCE: Dual-View Distribution Alignment for Dataset Condensation

TL;DR

The paper introduces DANCE, a novel distribution matching method for dataset condensation that leverages pre-trained models to improve synthetic data quality from both intra- and inter-class perspectives, achieving state-of-the-art results.

Contribution

DANCE innovatively combines inner-class and inter-class distribution alignment using pre-trained models to enhance dataset condensation efficiency and effectiveness.

Findings

Achieves state-of-the-art performance on dataset condensation tasks.

Maintains efficiency comparable to existing distribution-matching methods.

Effectively preserves class distributions during condensation.

Abstract

Dataset condensation addresses the problem of data burden by learning a small synthetic training set that preserves essential knowledge from the larger real training set. To date, the state-of-the-art (SOTA) results are often yielded by optimization-oriented methods, but their inefficiency hinders their application to realistic datasets. On the other hand, the Distribution-Matching (DM) methods show remarkable efficiency but sub-optimal results compared to optimization-oriented methods. In this paper, we reveal the limitations of current DM-based methods from the inner-class and inter-class views, i.e., Persistent Training and Distribution Shift. To address these problems, we propose a new DM-based method named Dual-view distribution AligNment for dataset CondEnsation (DANCE), which exploits a few pre-trained models to improve DM from both inner-class and inter-class views. Specifically, from the inner-class view, we construct multiple "middle encoders" to perform pseudo long-term distribution alignment, making the condensed set a good proxy of the real one during the whole training process; while from the inter-class view, we use the expert models to perform distribution calibration, ensuring the synthetic data remains in the real class region during condensing. Experiments demonstrate the proposed method achieves a SOTA performance while maintaining comparable efficiency with the original DM across various scenarios. Source codes are available at https://github.com/Hansong-Zhang/DANCE.