Privacy-Preserving Classification on Deep Neural Network

By Hervé Chabanne, Amaury de Wargny, Jonathan Milgram, Constance Morel, Emmanuel Prouff. 2017

281 citations recorded by the source corpus.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

1. M. Sadegh Riazi
2. Yang Liu
3. Hao Chen
4. Qinbin Li
5. Marcel Keller
6. Qiang Yang
7. Adrià Gascón
8. Assi Barak

Citations in the corpus

• Privacy in Deep Learning: A Survey (2020, methodology): “ARDEN [106] # # # MNIST, CIFAR-10, SVHN Image Classification w/ DNN Cloak [107] # # CIFAR-100, CelebA, UTKFace Image Classification w/ DNN Cryptonets [108] # # # MNIST Image Classification w/ DNN Private Classification [109] # # # MNIST Image Classification w/ DNN TAPAS [110] # # # MNIST, Faces, Can”
• Secure Evaluation of Quantized Neural Networks (2019, methodology): “[10] improve upon CryptoNets by evaluating networks with 6 hidden layers (as opposed to only 2 as in Gilad-Bachrach et al.”
• Federated Machine Learning (2019, background): “With the recent advances in deep learning, privacy-preserving neural networks inference is also receiving a lot of research interests[10, 11, 14, 28, 40, 52, 54].”
• CryptoDL: Deep Neural Networks over Encrypted Data (2017, methodology): “However, the degree of our polynomial is 3 while the polynomial used in [49] is of degree 2.”
• Chameleon: A Hybrid Secure Computation Framework for Machine Learning Applications (2018, background): “[27] also approximate the ReLu non-linear activation function using low-degree polynomials and provide a normalization layer prior to the activation layer.”
• Faster CryptoNets: Leveraging Sparsity for Real-World Encrypted Inference (2018, background): “Conveniently, batch norm transforms the pre-activation values into a normal distribution with zero mean and unit variance [41] which reduces overall error of the approximation [15].”
• Falcon: Honest-Majority Maliciously Secure Framework for Private Deep Learning (2020, methodology): “Batch normalization, has been previously considered in privacy-preserving inference as linear transformation using Homomorphic Encryption [22–24].”
• TAPAS: Tricks to Accelerate (encrypted) Prediction As a Service (2018, methodology): “(2016); Chabanne et al. (2017); Chen et al. (2017). Our approach is currently tailored to methods that have binary activations and we leave the implementation of these methods on encrypted data for future work.”
• CHET: Compiler and Runtime for Homomorphic Evaluation of Tensor Programs (2018, methodology): “[13] improve upon Cryptonets by using batch normalization [26] to bound the inputs to the activations into a small interval where the quadratic approximation is valid.”
• ABY3: A Mixed Protocol Framework for Machine Learning (2018, background): “One recent line of work uses fully homomorphic or somewhat homomorphic encryption to evaluate the model on encrypted data [34, 37, 19, 15].”
• Oblivious Neural Network Predictions via MiniONN Transformations (2017, background): “[16] also noticed the limited accuracy guarantees of the square function in CryptoNets.”
• Shredder: Learning Noise to Protect Privacy with Partial DNN Inference on the Edge (2019, methodology): “Secure multiparty computation (SMC) [16, 17] and homomorphic encryption [21, 19, 48] have also been used as attempts to deal with the privacy on offloaded computation on the cloud [19, 67, 68, 67, 48, 21, 52].”