Learning from Differentially Private Neural Activations with Edge Computing
Authors:
- Yunlong MaoYMYunlong Mao
- Shanhe YiSYShanhe Yi
- Qun LiQLQun Li
- Jinghao FengJFJinghao Feng
- Fengyuan XuFXFengyuan Xu
- Sheng ZhongSZSheng Zhong
- 1 October 2018
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 90-102
Abstract
Deep convolutional neural networks (DNNs) have brought significant performance improvements to face recognition. However the training can hardly be carried out on mobile devices because the training of these models requires much computational power. An individual user with the demand of deriving DNN models from her own datasets usually has to outsource the training procedure onto an edge server. However this outsourcing method violates privacy because it exposes the users' data to curious service providers. In this paper, we utilize the differentially private mechanism to enable the privacy-preserving edge based training of DNN face recognition models. During the training, DNN is split between the user device and the edge server in a way that both private data and model parameters are protected, with only a small cost of local computations. We rigorously prove that our approach is privacy preserving. We finally show that our mechanism is capable of training models in different scenarios, e.g., from scratch, or through fine-tuning over existed models.Keywords
This publication has 17 references indexed in Scilit:
- Deep Learning with Differential Privacy2016
- vDNN: Virtualized deep neural networks for scalable, memory-efficient neural network design2016
- Joint Face Detection and Alignment Using Multitask Cascaded Convolutional NetworksIEEE Signal Processing Letters, 2016
- Model Inversion Attacks that Exploit Confidence Information and Basic Countermeasures2015
- Privacy-Preserving Deep Learning2015
- A Survey of Fog Computing2015
- FaceNet: A unified embedding for face recognition and clustering2015
- Deep Face Recognition2015
- Deep Learning Face Representation from Predicting 10,000 Classes2014
- DeepFace: Closing the Gap to Human-Level Performance in Face Verification2014