Discrete Randomized Smoothing Meets Quantum Computing

TL;DR

This paper introduces a quantum computing approach to discrete randomized smoothing, significantly accelerating the certification process of ML models against adversarial attacks on discrete data.

Contribution

It combines quantum computing with discrete randomized smoothing, using superposition and Quantum Amplitude Estimation to improve certification efficiency for ML models.

Findings

Quadratic reduction in model calls using QAE

Effective certification on images, graphs, and text

New binary threat model for evaluation

Abstract

Breakthroughs in machine learning (ML) and advances in quantum computing (QC) drive the interdisciplinary field of quantum machine learning to new levels. However, due to the susceptibility of ML models to adversarial attacks, practical use raises safety-critical concerns. Existing Randomized Smoothing (RS) certification methods for classical machine learning models are computationally intensive. In this paper, we propose the combination of QC and the concept of discrete randomized smoothing to speed up the stochastic certification of ML models for discrete data. We show how to encode all the perturbations of the input binary data in superposition and use Quantum Amplitude Estimation (QAE) to obtain a quadratic reduction in the number of calls to the model that are required compared to traditional randomized smoothing techniques. In addition, we propose a new binary threat model to allow for an extensive evaluation of our approach on images, graphs, and text.