Accuracy-Privacy Trade-off in Deep Ensemble: A Membership Inference Perspective
Authors:
- 1 May 2023
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 364-381
Abstract
Deep ensemble learning has been shown to improve accuracy by training multiple neural networks and averaging their outputs. Ensemble learning has also been suggested to defend against membership inference attacks that undermine privacy. In this paper, we empirically demonstrate a trade-off between these two goals, namely accuracy and privacy (in terms of membership inference attacks), in deep ensembles. Using a wide range of datasets and model architectures, we show that the effectiveness of membership inference attacks increases when ensembling improves accuracy. We analyze the impact of various factors in deep ensembles and demonstrate the root cause of the trade-off. Then, we evaluate common defenses against membership inference attacks based on regularization and differential privacy. We show that while these defenses can mitigate the effectiveness of membership inference attacks, they simultaneously degrade ensemble accuracy. We illustrate similar trade-off in more advanced and state-of-the-art ensembling techniques, such as snapshot ensembles and diversified ensemble networks. Finally, we propose a simple yet effective defense for deep ensembles to break the trade-off and, consequently, improve the accuracy and privacy, simultaneously.Keywords
All Related Versions
This publication has 24 references indexed in Scilit:
- Machine Learning with Membership Privacy using Adversarial Regularization2018
- Privacy Risk in Machine Learning: Analyzing the Connection to Overfitting2018
- Ensemble learning: A surveyWIREs Data Mining and Knowledge Discovery, 2018
- Membership Inference Attacks Against Machine Learning Models2017
- Model Inversion Attacks that Exploit Confidence Information and Basic Countermeasures2015
- ImageNet Large Scale Visual Recognition ChallengeInternational Journal of Computer Vision, 2015
- Hacking smart machines with smarter ones: How to extract meaningful data from machine learning classifiersInternational Journal of Security and Networks, 2015
- Differential Privacy: A Survey of Results2008
- Measures of Diversity in Classifier Ensembles and Their Relationship with the Ensemble AccuracyMachine Learning, 2003
- Gradient-based learning applied to document recognitionProceedings of the IEEE, 1998