Effectiveness of Arbitrary Transfer Sets for Data-free Knowledge Distillation

TL;DR

This paper demonstrates that arbitrary, unrelated datasets can effectively be used for data-free knowledge distillation if they are balanced across target classes, simplifying the process compared to complex synthetic data methods.

Contribution

It introduces a simple approach using arbitrary datasets for data-free knowledge distillation and validates its effectiveness across multiple benchmark datasets.

Findings

Arbitrary datasets can effectively replace original training data in knowledge distillation.

Balanced class distribution in the transfer set is crucial for success.

The approach simplifies data-free distillation without complex generative models.

Abstract

Knowledge Distillation is an effective method to transfer the learning across deep neural networks. Typically, the dataset originally used for training the Teacher model is chosen as the "Transfer Set" to conduct the knowledge transfer to the Student. However, this original training data may not always be freely available due to privacy or sensitivity concerns. In such scenarios, existing approaches either iteratively compose a synthetic set representative of the original training dataset, one sample at a time or learn a generative model to compose such a transfer set. However, both these approaches involve complex optimization (GAN training or several backpropagation steps to synthesize one sample) and are often computationally expensive. In this paper, as a simple alternative, we investigate the effectiveness of "arbitrary transfer sets" such as random noise, publicly available synthetic, and natural datasets, all of which are completely unrelated to the original training dataset in terms of their visual or semantic contents. Through extensive experiments on multiple benchmark datasets such as MNIST, FMNIST, CIFAR-10 and CIFAR-100, we discover and validate surprising effectiveness of using arbitrary data to conduct knowledge distillation when this dataset is "target-class balanced". We believe that this important observation can potentially lead to designing baselines for the data-free knowledge distillation task.