Adversarial Robustness of Partitioned Quantum Classifiers

TL;DR

This paper investigates the adversarial robustness of partitioned quantum classifiers, analyzing how circuit cutting and teleportation techniques affect their vulnerability to adversarial attacks, with both theoretical and experimental insights.

Contribution

It introduces a study of adversarial robustness in partitioned quantum classifiers, linking perturbations to adversarial gates and providing theoretical and experimental analysis.

Findings

Perturbations in wire cutting and teleportation impact quantum classifier robustness.

A connection between adversarial perturbations and adversarial gates is established.

Both theoretical and experimental results demonstrate vulnerabilities and potential defenses.

Abstract

Adversarial robustness in quantum classifiers is a critical area of study, providing insights into their performance compared to classical models and uncovering potential advantages inherent to quantum machine learning. In the NISQ era of quantum computing, circuit cutting is a notable technique for simulating circuits that exceed the qubit limitations of current devices, enabling the distribution of a quantum circuit's execution across multiple quantum processing units through classical communication. In contrast, when quantum communication is available, teleportation-based methods can be used to support the distribution of the quantum circuit. We study the robustness of partitioned quantum classifiers to adversarial perturbations targeting wire cutting or quantum state teleportation and show a link between such perturbations and implementing adversarial gates within intermediate layers of a quantum classifier. We then proceed to study the latter problem from both a theoretical and experimental perspective.