An Extension of Fano's Inequality for Characterizing Model Susceptibility to Membership Inference Attacks
arXiv:2009.08097
Abstract
Deep neural networks have been shown to be vulnerable to membership inference attacks wherein the attacker aims to detect whether specific input data were used to train the model. These attacks can potentially leak private or proprietary data. We present a new extension of Fano's inequality and employ it to theoretically establish that the probability of success for a membership inference attack on a deep neural network can be bounded using the mutual information between its inputs and its activations. This enables the use of mutual information to measure the susceptibility of a DNN model to membership inference attacks. In our empirical evaluation, we show that the correlation between the mutual information and the susceptibility of the DNN model to membership inference attacks is 0.966, 0.996, and 0.955 for CIFAR-10, SVHN and GTSRB models, respectively.
arXiv comment9 pages, 3 figures
References in corpus (13)
- Deep Learning with Differential Privacy
- Comprehensive Privacy Analysis of Deep Learning: Passive and Active White-box Inference Attacks against Centralized and Federated Learning
- Understanding deep learning requires rethinking generalization
- Opening the Black Box of Deep Neural Networks via Information
- Scalable Private Learning with PATE
- Differential Privacy as a Mutual Information Constraint
- Understanding Membership Inferences on Well-Generalized Learning Models
- Towards Demystifying Membership Inference Attacks
- Efficient Deep Learning on Multi-Source Private Data
- Privacy-preserving Machine Learning through Data Obfuscation
- Towards Measuring Membership Privacy
- Where is the Information in a Deep Neural Network?
- Privacy-preserving Prediction