Radioactive data: tracing through training
Published in
International Conference on Machine Learning(2020)External Links:
TL
TL;DR
A new technique is proposed that makes imperceptible changes to this dataset such that any model trained on it will bear an identifiable mark, robust to data augmentation and the stochasticity of deep network optimization.
Abstract
We want to detect whether a particular image dataset has been used to train a model. We propose a new technique, \emph{radioactive data}, that makes imperceptible changes to this dataset such that any model trained on it will bear an identifiable mark. The mark is robust to strong variations such as different architectures or optimization methods. Given a trained model, our technique detects the use of radioactive data and provides a level of confidence (p-value). Our experiments on large-scale benchmarks (Imagenet), using standard architectures (Resnet-18, VGG-16, Densenet-121) and training procedures, show that we can detect usage of radioactive data with high confidence (p<10^-4) even when only 1% of the data used to trained our model is radioactive. Our method is robust to data augmentation and the stochasticity of deep network optimization. As a result, it offers a much higher signal-to-noise ratio than data poisoning and backdoor methods.
References47 items
1
White-box vs Black-box: Bayes Optimal Strategies for Membership Inference
2
Unsupervised Pre-Training of Image Features on Non-Curated Data
3
BadNets: Evaluating Backdooring Attacks on Deep Neural Networks
4
Are Deep Neural Networks good for blind image watermarking?
5