Dataset Distillation via Adversarial Prediction Matching
This work addresses the problem of efficiently condensing large datasets for machine learning practitioners, offering a memory-efficient method with significant performance gains, though it is incremental in improving existing distillation techniques.
The paper tackles dataset distillation by minimizing prediction discrepancies between models trained on original and distilled datasets using an adversarial framework, achieving 94% of the test accuracy with datasets 10% the size of the original, including on ImageNet-1K, while using 2.5× less memory and 5× less runtime than state-of-the-art methods.
Dataset distillation is the technique of synthesizing smaller condensed datasets from large original datasets while retaining necessary information to persist the effect. In this paper, we approach the dataset distillation problem from a novel perspective: we regard minimizing the prediction discrepancy on the real data distribution between models, which are respectively trained on the large original dataset and on the small distilled dataset, as a conduit for condensing information from the raw data into the distilled version. An adversarial framework is proposed to solve the problem efficiently. In contrast to existing distillation methods involving nested optimization or long-range gradient unrolling, our approach hinges on single-level optimization. This ensures the memory efficiency of our method and provides a flexible tradeoff between time and memory budgets, allowing us to distil ImageNet-1K using a minimum of only 6.5GB of GPU memory. Under the optimal tradeoff strategy, it requires only 2.5$\times$ less memory and 5$\times$ less runtime compared to the state-of-the-art. Empirically, our method can produce synthetic datasets just 10% the size of the original, yet achieve, on average, 94% of the test accuracy of models trained on the full original datasets including ImageNet-1K, significantly surpassing state-of-the-art. Additionally, extensive tests reveal that our distilled datasets excel in cross-architecture generalization capabilities.