Compilation as a Defense: Enhancing DL Model Attack Robustness via Tensor Optimization

TL;DR

This paper proposes a novel defense against adversarial machine learning side-channel attacks by using tensor optimization during model compilation, significantly reducing attack effectiveness.

Contribution

It introduces tensor optimization as an effective compilation-based defense mechanism against AML side-channel attacks, addressing a gap in efficient remediation strategies.

Findings

Model attack effectiveness decreases by up to 43% with tensor optimization.

Tensor optimization offers a promising direction for AML attack mitigation.

The approach avoids costly model re-engineering.

Abstract

Adversarial Machine Learning (AML) is a rapidly growing field of security research, with an often overlooked area being model attacks through side-channels. Previous works show such attacks to be serious threats, though little progress has been made on efficient remediation strategies that avoid costly model re-engineering. This work demonstrates a new defense against AML side-channel attacks using model compilation techniques, namely tensor optimization. We show relative model attack effectiveness decreases of up to 43% using tensor optimization, discuss the implications, and direction of future work.