Self-Ensemble Protection: Training Checkpoints Are Good Data Protectors

TL;DR

This paper introduces Self-Ensemble Protection (SEP), a novel method that uses model checkpoints' gradients to generate imperceptible perturbations, effectively protecting data from being used to train high-performance models.

Contribution

The paper proposes a new data protection method, SEP, leveraging checkpoints' gradients to create robust, imperceptible perturbations that hinder model training on protected data.

Findings

SEP achieves state-of-the-art protection performance.

Perturbations significantly reduce model accuracy on protected data.

Checkpoints' gradients are diverse, enabling effective ensemble-based protection.

Abstract

As data becomes increasingly vital, a company would be very cautious about releasing data, because the competitors could use it to train high-performance models, thereby posing a tremendous threat to the company's commercial competence. To prevent training good models on the data, we could add imperceptible perturbations to it. Since such perturbations aim at hurting the entire training process, they should reflect the vulnerability of DNN training, rather than that of a single model. Based on this new idea, we seek perturbed examples that are always unrecognized (never correctly classified) in training. In this paper, we uncover them by model checkpoints' gradients, forming the proposed self-ensemble protection (SEP), which is very effective because (1) learning on examples ignored during normal training tends to yield DNNs ignoring normal examples; (2) checkpoints' cross-model gradients are close to orthogonal, meaning that they are as diverse as DNNs with different architectures. That is, our amazing performance of ensemble only requires the computation of training one model. By extensive experiments with 9 baselines on 3 datasets and 5 architectures, SEP is verified to be a new state-of-the-art, e.g., our small $\ell_\infty=2/255$ perturbations reduce the accuracy of a CIFAR-10 ResNet18 from 94.56% to 14.68%, compared to 41.35% by the best-known method. Code is available at https://github.com/Sizhe-Chen/SEP.