Distilled Pruning: Using Synthetic Data to Win the Lottery
This addresses resource-efficient model compression for deep learning practitioners, though it appears incremental as it builds on existing pruning methods like Lottery Tickets.
The paper tackles the problem of pruning deep learning models by using distilled synthetic data to find sparse, trainable subnetworks, achieving up to 5x faster pruning than Iterative Magnitude Pruning on CIFAR-10 at comparable sparsity.
This work introduces a novel approach to pruning deep learning models by using distilled data. Unlike conventional strategies which primarily focus on architectural or algorithmic optimization, our method reconsiders the role of data in these scenarios. Distilled datasets capture essential patterns from larger datasets, and we demonstrate how to leverage this capability to enable a computationally efficient pruning process. Our approach can find sparse, trainable subnetworks (a.k.a. Lottery Tickets) up to 5x faster than Iterative Magnitude Pruning at comparable sparsity on CIFAR-10. The experimental results highlight the potential of using distilled data for resource-efficient neural network pruning, model compression, and neural architecture search.