Calibrated Dataset Condensation for Faster Hyperparameter Search

TL;DR

This paper introduces a hyperparameter-calibrated dataset condensation method that creates synthetic validation datasets to preserve model ranking consistency, significantly accelerating hyperparameter and architecture search on images and graphs.

Contribution

The paper proposes a novel HCDC algorithm that matches hyperparameter gradients for better generalization in dataset condensation, improving hyperparameter search efficiency.

Findings

Maintains validation-performance rankings across models and hyperparameters.

Speeds up hyperparameter and architecture search.

Effective on both image and graph datasets.

Abstract

Dataset condensation can be used to reduce the computational cost of training multiple models on a large dataset by condensing the training dataset into a small synthetic set. State-of-the-art approaches rely on matching the model gradients between the real and synthetic data. However, there is no theoretical guarantee of the generalizability of the condensed data: data condensation often generalizes poorly across hyperparameters/architectures in practice. This paper considers a different condensation objective specifically geared toward hyperparameter search. We aim to generate a synthetic validation dataset so that the validation-performance rankings of the models, with different hyperparameters, on the condensed and original datasets are comparable. We propose a novel hyperparameter-calibrated dataset condensation (HCDC) algorithm, which obtains the synthetic validation dataset by matching the hyperparameter gradients computed via implicit differentiation and efficient inverse Hessian approximation. Experiments demonstrate that the proposed framework effectively maintains the validation-performance rankings of models and speeds up hyperparameter/architecture search for tasks on both images and graphs.