Differential Privacy Preservation for Deep Auto-Encoders: an Application of Human Behavior Prediction

Abstract

In recent years, deep learning has spread beyond both academia and industry with many exciting real-world applications. The development of deep learning has presented obvious privacy issues. However, there has been lack of scientific study about privacy preservation in deep learning. In this paper, we concentrate on the auto-encoder, a fundamental component in deep learning, and propose the deep private auto-encoder (dPA). Our main idea is to enforce ε-differential privacy by perturbing the objective functions of the traditional deep auto-encoder, rather than its results. We apply the dPA to human behavior prediction in a health social network. Theoretical analysis and thorough experimental evaluations show that the dPA is highly effective and efficient, and it significantly outperforms existing solutions.

Differential Private Deep Learning Models for Analyzing Breast Cancer Omics Data

“…Finally, we used these 2000-size DP representations of PD 1 patients to build our final dpClassM model. For the TCGA dataset, Figures 3A, B shows the comparison (mean accuracy and AUC of 10-fold cross-validation) of dpClassM against the baselines (25,31,32) with the same privacy budget (∈ = 1.0). In Figures 3A, B, the x-axis represents the ER status (ER+/-) and eight pairs of cancer types (number cases of the Supplementary Table S1) that we choose to perform our experiments.…”

“…They used FM(( 24)) to perturb the objective function's coefficients to build DP-DNN. However, FM(( 24)) follows ∈-DP which may affect the performance of Phan et al (32) framework in many real life applications (29).…”

“…Phan et al. ( 32 ) was the first work to build a DP-DNN autoencoder that can provide state-of-the-art regression performances such as the prediction of human behaviors from health social networks. They used FM((24)) to perturb the objective function’s coefficients to build DP-DNN.…”

“…However, FM((24)) follows ∈-DP which may affect the performance of Phan et al. ( 32 ) framework in many real life applications ( 29 ).…”

Deep learning for privacy preservation in autonomous moving platforms enhanced 5G heterogeneous networks

“…Using differential privacy technology can effectively protect users' privacy when they provide information to ISPs or service providers (SPs). The effect of differential privacy can also be enhanced with DL technologies [152], [153]. In [154], the authors proposed a new method for learning and a refined analysis of privacy costs within the framework of differential privacy.…”