Soft-Label Dataset Distillation and Text Dataset Distillation

By Ilia Sucholutsky, Matthias Schonlau. 2019

162 citations recorded by the source corpus.

Dataset distillation is a method for reducing dataset sizes by learning a small number of representative synthetic samples. This has several benefits such as speeding up model training, reducing energy consumption, and reducing required storage space. These benefits are especially crucial in settings like federated learning where initial overhead costs are justified by the speedup they enable. Currently, 1) each synthetic sample is assigned a single ‘hard’ label, and 2) dataset distillation can only be used with image data. We propose to simultaneously distill both images and their labels, thus assigning each synthetic sample a ‘soft’ label (a distribution of labels). Our algorithm increases accuracy by 2-4% for several image classification tasks. Using ‘soft’ labels also enables distilled datasets to consist of fewer samples than there are classes as each sample encodes information for multiple classes. For example, training a LeNet model with 10 distilled images (one per class) results in over 96% accuracy on MNIST, and almost 92% accuracy when trained on just 5 distilled images. We also extend the dataset distillation algorithm to distill text data. We demonstrate that text distillation outperforms other methods across multiple datasets. For example, models attain almost their original accuracy on the IMDB sentiment analysis task using just 20 distilled sentences. Our code can be found at https://github.com/ilia10000/dataset-distillation.

Subfields: Artificial Intelligence

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

Citations in the corpus

Distilled One-Shot Federated Learning (2020, methodology): “To achieve good results even when client data is poorly distributed, we leverage soft labels [30] and introduce two new techniques: soft reset and random masking.”
Dataset Condensation with Differentiable Siamese Augmentation (2021, background): “Sucholutsky & Schonlau (2019) extend DD by learning synthetic images and soft labels simultaneously.”
Flexible Dataset Distillation: Learn Labels Instead of Images (2020, methodology): “Dataset distillation Most closely related to our work is Dataset [31] and Soft-Label Dataset Distillation [27].”
Dataset Distillation by Matching Training Trajectories (2022, methodology): “Dataset distillation has become a lively research topic in machine learning [2, 25, 26, 38, 45, 46, 47] with various applications, such as continual learning, neural architecture search, and privacy-preserving ML.”
Dataset Condensation with Distribution Matching (2021, methodology): “Previous works [45, 40, 39, 6, 31, 32] formulate the dataset condensation as a learning-to-learn problem, pose the network parameters θS as a function of synthetic data S and obtain a solution for S by minimizing the training loss LT over the original data T :”
Dataset Meta-Learning from Kernel Ridge-Regression (2020, background): “The works Sucholutsky & Schonlau (2019); Bohdal et al. (2020) build upon Wang et al. (2018) by distilling labels.”