MixNN: A design for protecting deep learning models
Not on the map. Why?
Abstract (arXiv)
In this paper, we propose a novel design, called MixNN, for protecting deep learning model structure and parameters. The layers in a deep learning model of MixNN are fully decentralized. It hides communication address, layer parameters and operations, and forward as well as backward message flows among non-adjacent layers using the ideas from mix networks. MixNN has following advantages: 1) an adversary cannot fully control all layers of a model including the structure and parameters, 2) even some layers may collude but they cannot tamper with other honest layers, 3) model privacy is preserved in the training phase. We provide detailed descriptions for deployment. In one classification experiment, we compared a neural network deployed in a virtual machine with the same one using the MixNN design on the AWS EC2. The result shows that our MixNN retains less than 0.001 difference in terms of classification accuracy, while the whole running time of MixNN is about 7.5 times slower than the one running on a single virtual machine.
Key facts. No citing papers in the map yet (data through 7 Oct 2026). Builds on 16 of its 39 references that are in the map. First posted on arXiv 28 Mar 2022.
Builds on
16 of 39 references.
References · 16
- 2014Very Deep Convolutional Networks for Large-Scale Image Recognitioncited by 15,267
- 2016Membership Inference Attacks against Machine Learning Modelscited by 2,221
- 2016Stealing Machine Learning Models via Prediction APIscited by 602
- 2019Local Model Poisoning Attacks to Byzantine-Robust Federated Learningcited by 479
- 2019Federated Machine Learning: Concept and Applicationscited by 409
- 2018Split learning for health: Distributed deep learning without sharing raw patient datacited by 391
- 2020When Machine Learning Meets Privacy: A Survey and Outlookcited by 67
- 2020DarkneTZ: Towards Model Privacy at the Edge using Trusted Execution Environmentscited by 57
- 2018Cache Telepathy: Leveraging Shared Resource Attacks to Learn DNN Architecturescited by 56
- 2017DeepSecure: Scalable Provably-Secure Deep Learningcited by 48
- 2018Faster CryptoNets: Leveraging Sparsity for Real-World Encrypted Inferencecited by 45
- 2014Crypto-Nets: Neural Networks over Encrypted Datacited by 10
4 more
- 2017Dropping Activation Outputs with Localized First-layer Deep Network for Enhancing User Privacy and Data Securitycited by 8
- 2020Privacy-Preserving Deep Learning Computation for Geo-Distributed Medical Big-Data Platformscited by 6
- 2017The Loopix Anonymity Systemcited by 3
- 2016Atom: Horizontally Scaling Strong Anonymitycited by 2
Built on by
Not cited yet.