Privacy and Integrity Preserving Training Using Trusted Hardware

TL;DR

The paper introduces DarKnight, a framework that combines trusted hardware and accelerators to enable privacy-preserving and integrity-verified training of large deep neural networks in cloud environments.

Contribution

It proposes a novel cooperative execution model using trusted execution environments and accelerators for secure DNN training.

Findings

Ensures input privacy during training in cloud settings

Verifies computation integrity using trusted hardware

Supports large DNN training with hardware acceleration

Abstract

Privacy and security-related concerns are growing as machine learning reaches diverse application domains. The data holders want to train with private data while exploiting accelerators, such as GPUs, that are hosted in the cloud. However, Cloud systems are vulnerable to attackers that compromise the privacy of data and integrity of computations. This work presents DarKnight, a framework for large DNN training while protecting input privacy and computation integrity. DarKnight relies on cooperative execution between trusted execution environments (TEE) and accelerators, where the TEE provides privacy and integrity verification, while accelerators perform the computation heavy linear algebraic operations.