Privacy-Preserving Deep Learning via Additively Homomorphic Encryption

By Le Trieu Phong, Yoshinori Aono, Takuya Hayashi, Lihua Wang, Shiho Moriai. 2018

1,671 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Privacy & Security

How researchers see this paper

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

In Artificial Intelligence

In Privacy & Security

Citations in the corpus

NoPeek: Information leakage reduction to share activations in distributed deep learning (2020, neutral): “The work in [39, 40, 41] describes HE and garbled circuits (an MPC scheme) for privacy-preserving biometric identification.”
Backdoor Attacks on Federated Meta-Learning (2020, background): “Additional privacy threats have also been addressed in federated learning: for example, secure aggregation techniques have been proposed where users send encrypted updates that the server applies to an encrypted model [2, 3].”
Privacy in Deep Learning: A Survey (2020, methodology): “w/ DNN Distributed DP [87] # # eICU, TCGA Classification w/ DNN DP model publishing [88] # # MNIST, CIFAR Image Classification w/ DNN DP federated learning [89] # # MNIST Image Classification w/ DNN ScalarDP, PrivUnit [90] # # MNIST, CIFAR Image Classification w/ DNN DSSGD [21] # # MNIST, SVHN Image”
Central Server Free Federated Learning over Single-sided Trust Social Networks (2019, background): “push-sum (OPS), which possess the following features: Only models rather than local gradient are exchanged among clients in our algorithm. This can reduce the risk of exposing clients’ data privacy (Aono et al., 2017). Our algorithm removes some constraints imposed by typical decentralized methods, ”
Federated Machine Learning (2019, methodology): “al model parameters. Security Analysis. The above architecture is proved to protect data leakage against the semihonest server, if gradients aggregation is done with SMC [9] or Homomorphic Encryption [51]. But it may be subject to attack in another security model by a malicious participant training ”
On Safeguarding Privacy and Security in the Framework of Federated Learning (2019, background): “In addition, FL can expose intermediate results such as parameter updates from an optimization algorithm like stochastic gradient descent (SGD), and the transmission of these gradients may actually leak private information [9] when exposed together with a data structure such as image pixels.”
The Value of Collaboration in Convex Machine Learning with Differential Privacy (2019, methodology): “Training ML models using encrypted data was discussed in [10]–[14].”
Chameleon: A Hybrid Secure Computation Framework for Machine Learning Applications (2018, background): “Phong et al. [61] propose a mechanism for privacy-preserving deep learning based on additively homomorphic encryption.”
Privacy‐preserving federated learning based on multi‐key homomorphic encryption (2021, methodology): “Improved schemes have been proposed in [38], [7] using addictive homomorphic encryption to protect model updates, by Zhang et al. [8] exploiting privacy-preserving and veriﬁable federated learning using the Paillier cryptosystem [10], as well as by Froelicher et al. [39] who proposed a…”
Privacy-Preserving Inference in Machine Learning Services Using Trusted Execution Environments (2019, background): “Recent works [2], [16], [18], [29], [31] have explored using this technique to provide secure inference on models in the setting of adversaries trying to learn information about the data.”
No Peek: A Survey of private distributed deep learning (2018, methodology): “The works in [17,19] apply these ideas in the context of deep learning.”
When Machine Learning Meets Privacy (2020, methodology): “. ACM Comput. Surv., Vol. 1, No. 1, Article . Publication date: November 2020. 12 Bo Liu, et al. •Encrypting ML model. The encryption technique is also used to protect the model privacy. Phong et al. [126] proposed to use “additively homomorphic encryption on the gradients”. The scheme can prevent i”
Efficient and Privacy-Enhanced Federated Learning for Industrial Artificial Intelligence (2020, background): “[19] proposed a collaborativeDLmodel via additively HE.”
Federated Learning on Non-IID Data: A Survey (2021, methodology): “Even worse, DP cannot deal with some inverting gradient attacks as introduced in [107, 51].”
PrivEdge: From Local to Distributed Private Training and Prediction (2020, methodology): “honest in performing the operations, but also attempting to learn information about other users by analysing the received information [16]) or by an honest-but-curious service provider during training or prediction [13], [17]–[19].”
Security and Privacy Issues in Deep Learning (2018, methodology): “Against such privacy threats, privacy-preserving techniques including fully homomorphic encryption (FHE) [16, 17, 18, 19, 20], differential privacy [21, 22, 23, 24, 25, 26, 27, 28], and secure multi-party computation (SMC) [29, 30], have been combined with the DL frameworks.”