Mining Data Impressions from Deep Models as Substitute for the Unavailable Training Data

TL;DR

This paper introduces 'Data Impressions', synthetic data extracted from pretrained deep models' parameters, enabling various tasks like domain adaptation, continual learning, and adversarial robustness without access to original training data.

Contribution

It proposes a novel method to generate proxy training data from model parameters, facilitating multiple applications in data privacy scenarios.

Findings

Effective in unsupervised domain adaptation

Enhances continual learning performance

Generates data-free universal adversarial perturbations

Abstract

Pretrained deep models hold their learnt knowledge in the form of model parameters. These parameters act as "memory" for the trained models and help them generalize well on unseen data. However, in absence of training data, the utility of a trained model is merely limited to either inference or better initialization towards a target task. In this paper, we go further and extract synthetic data by leveraging the learnt model parameters. We dub them "Data Impressions", which act as proxy to the training data and can be used to realize a variety of tasks. These are useful in scenarios where only the pretrained models are available and the training data is not shared (e.g., due to privacy or sensitivity concerns). We show the applicability of data impressions in solving several computer vision tasks such as unsupervised domain adaptation, continual learning as well as knowledge distillation. We also study the adversarial robustness of lightweight models trained via knowledge distillation using these data impressions. Further, we demonstrate the efficacy of data impressions in generating data-free Universal Adversarial Perturbations (UAPs) with better fooling rates. Extensive experiments performed on benchmark datasets demonstrate competitive performance achieved using data impressions in absence of original training data.