Dataset Condensation with Distribution Matching
Preprint
Abstract
Computational cost of training state-of-the-art deep models in many learning problems is rapidly increasing due to more sophisticated models and larger datasets. A recent promising direction for reducing training cost is dataset condensation that aims to replace the original large training set with a significantly smaller learned synthetic set while preserving the original information. While training deep models on the small set of condensed images can be extremely fast, their synthesis remains computationally expensive due to the complex bi-level optimization and second-order derivative computation. In this work, we propose a simple yet effective method that synthesizes condensed images by matching feature distributions of the synthetic and original training images in many sampled embedding spaces. Our method significantly reduces the synthesis cost while achieving comparable or better performance. Thanks to its efficiency, we apply our method to more realistic and larger datasets with sophisticated neural architectures and obtain a significant performance boost. We also show promising practical benefits of our method in continual learning and neural architecture search.Keywords
All Related Versions
This publication has 6 references indexed in Scilit:
- CAFE: Learning to Condense Dataset by Aligning Features2022
- Soft-Label Dataset Distillation and Text Dataset Distillation2021
- A review on neural networks with random weightsNeurocomputing, 2018
- Deep Neural Networks with Random Gaussian Weights: A Universal Classification Strategy?IEEE Transactions on Signal Processing, 2016
- Querying Discriminative and Representative Samples for Batch Mode Active LearningACM Transactions on Knowledge Discovery From Data, 2015
- Gradient-based learning applied to document recognitionProceedings of the IEEE, 1998