Deep Learning on Private Data

By M. Sadegh Riazi, Bita Darvish Rouani, Farinaz Koushanfar. 2019

41 citations recorded by the source corpus.

Emerging complex deep neural networks require vast amounts of data to achieve high precision. However, the information is often collected from user logs and personal data. In this article, we summarize recent cryptographic methodologies for provably privacy-preserving deep learning and inference.

Subfields: Artificial Intelligence

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

Citations in the corpus

Mitigating Sybils in Federated Learning Poisoning (2018, background): “Federated learning also enables model training across a set of clients with highly private data [30].”
A Roadmap Toward the Resilient Internet of Things for Cyber-Physical Systems (2018, background): “A common approach is to encrypt the data or underlying model [190]–[192].”
XONN: XNOR-based Oblivious Deep Neural Network Inference (2019, background): “An overview of re-lated frameworks is provided in [47, 48].”
HEAX: High-Performance Architecture for Computation on Homomorphically Encrypted Data in the Cloud (2019, background): “Privacy-preserving MLaaS has been studied rigorously in recent years and several protocol-level and algorithmic improvements have been proposed [36, 48, 63, 46, 19, 64, 62, 61, 42].”
DarKnight: A Data Privacy Scheme for Training and Inference of Deep Neural Networks (2020, background): “Adversary capabilities: While adversaries can perform various attacks on DNN models and datasets (Riazi et al., 2019), DarKnight focuses on attacks that expose the datasets used in training or inference and attacks that modify computational results on untrusted hardware.”