Bilateral Dependency Optimization: Defending Against Model-inversion Attacks

Abstract

Through using only a well-trained classifier, model-inversion (MI) attacks can recover the data used for training the classifier, leading to the privacy leakage of the training data. To defend against MI attacks, previous work utilizes a unilateral dependency optimization strategy, i.e., minimizing the dependency between inputs (i.e., features) and outputs (i.e., labels) during training the classifier. However, such a minimization process conflicts with minimizing the supervised loss that aims to maximize the dependency between inputs and outputs, causing an explicit trade-off between model robustness against MI attacks and model utility on classification tasks. In this paper, we aim to minimize the dependency between the latent representations and the inputs while maximizing the dependency between latent representations and the outputs, named a bilateral dependency optimization (BiDO) strategy. In particular, we use the dependency constraints as a universally applicable regularizer in addition to commonly used losses for deep neural networks (e.g., cross-entropy), which can be instantiated with appropriate dependency criteria according to different tasks. To verify the efficacy of our strategy, we propose two implementations of BiDO, by using two different dependency measures: BiDO with constrained covariance (BiDO-COCO) and BiDO with Hilbert-Schmidt Independence Criterion (BiDO-HSIC). Experiments show that BiDO achieves the state-of-the-art defense performance for a variety of datasets, classifiers, and MI attacks while suffering a minor classification-accuracy drop compared to the well-trained classifier with no defense, which lights up a novel road to defend against MI attacks.

arXiv commentAccepted to KDD 2022 (Research Track)

References in corpus (10)

• Adam: A Method for Stochastic Optimization
• Very Deep Convolutional Networks for Large-Scale Image Recognition
• Dipole: Diagnosis Prediction in Healthcare via Attention-based Bidirectional Recurrent Neural Networks
• Evaluating Gradient Inversion Attacks and Defenses in Federated Learning
• Thieves on Sesame Street! Model Extraction of BERT-based APIs
• Updates-Leak: Data Set Inference and Reconstruction Attacks in Online Learning
• Learning Deep Kernels for Non-Parametric Two-Sample Tests
• Defending Model Inversion and Membership Inference Attacks via Prediction Purification
• Variational Model Inversion Attacks
• Self-Supervised Learning with Kernel Dependence Maximization

Cited by in corpus (1)

• Unstoppable Attack: Label-Only Model Inversion via Conditional Diffusion Model