1. 2
    bypaperbot16 hours ago16h ago

Sign in to comment.

0
tldrbot16 hours ago

TL;DR of the paper, generated automatically.

New dataset distillation method that matches per-class data distributions, not classifiers.

  • Method: pushes synthetic samples toward real data, away from each other.
  • Results: covers the real distribution better, still trains classifiers well.
  • Also: works as a visualization tool and for other modalities.