Distilling Datasets Into Less Than One Image

arXiv:2403.12040v1 Announce Type: new
Abstract: Dataset distillation aims to compress a dataset into a much smaller one so that a model trained on the distilled dataset achieves high accuracy. Current methods frame this as maximizing the distilled classification accuracy for a budget of K distilled images-per-class, where K is a positive integer. In this paper, we push the boundaries of dataset distillation, compressing the dataset into less than an image-per-class. It is important to realize that the meaningful quantity is …

abstract accuracy arxiv budget class classification cs.cv current dataset datasets distillation image images per positive type