Le Trieu Phong

Scalable and Secure Logistic Regression via Homomorphic Encryption

Logistic regression is a powerful machine learning tool to classify data. When dealing with sensitive data such as private or medical information, cares are necessary. In this paper, we propose a secure system for protecting the training data in logistic regression via homomorphic encryption. Perhaps surprisingly, despite the non-polynomial tasks of training in logistic regression, we show that only additively homomorphic encryption is needed to build our system. Our system is secure and scalable with the dataset size.

Privacy-Preserving Deep Learning: Revisited and Enhanced

We build a privacy-preserving deep learning system in which many learning participants perform neural network-based deep learning over a combined dataset of all, without actually revealing the participants’ local data to a curious server. To that end, we revisit the previous work by Shokri and Shmatikov (ACM CCS 2015) and point out that local data information may be actually leaked to an honest-but-curious server. We then move on to fix that problem via building an enhanced system with following properties: (1) no information is leaked to the server; and (2) accuracy is kept intact, compared to that of the ordinary deep learning system also over the combined dataset. Our system makes use of additively homomorphic encryption, and we show that our usage of encryption adds little overhead to the ordinary deep learning system.

Efficient Key-Rotatable and Security-Updatable Homomorphic Encryption

In this paper we presents the notion of key-rotatable and security-updatable homomorphic encryption (KR-SU-HE) scheme, which is a class of public-key homomorphic encryption in which the keys and the security of any ciphertext can be rotated and updated while still keeping the underlying plaintext intact and unrevealed. We formalise syntax and security notions for KR-SU-HE schemes and then build a concrete scheme based on the Learning With Errors assumption. We then perform testing implementation to show that our proposed scheme is efficiently practical.

A Generic yet Efficient Method for Secure Inner Product

Secure inner product, namely the computation of inner product whose terms are all in encrypted form, is the central technique for various privacy-preserving applications. In this paper, we propose a generic yet efficient method to compute secure inner products of vectors (or matrices) using matrix trace properties. Indeed, our method not only applies to both LWE-based and ring-LWE-based homomorphic encryption schemes, but also is more efficient compared to previously known methods.

Privacy-Preserving Stochastic Gradient Descent with Multiple Distributed Trainers

Assume that there are L local datasets distributed among L owners (also called trainers hereafter). The problem is as follows: the owners wish to apply a machine learning method over the combined dataset of all to obtain the best possible learning output; but do not want to publicly share the local datasets due to privacy concerns. In this paper we design a system solving the problem in which stochastic gradient descent (SGD) algorithm is used as the machine learning method, as SGD is at the heart of recent deep learning techniques. Our system differs from existing work by following features: (1) we do not share the gradients in SGD but share the weight parameters; and (2) we use symmetric encryption to protect the weight parameters against an honest-but-curious server used as a common place for storage. Therefore, we are able to avoid information leakage of local data to the server; and the efficiency of our system is kept reasonably compared to the original SGD over the combined dataset. Finally, we experiment over a real dataset to verify the practicality of our system.