Quantum Support Vector Regression for Robust Anomaly Detection

TL;DR

This paper evaluates Quantum Support Vector Regression for anomaly detection on IBM quantum hardware, demonstrating its robustness to certain noise types but vulnerability to adversarial attacks.

Contribution

It provides a comprehensive benchmark of QSVR's performance on real quantum hardware and analyzes its robustness and vulnerabilities in noisy and adversarial settings.

Findings

QSVR outperforms noiseless simulation on some datasets.

Model is robust to depolarizing, phase damping, phase flip, and bit flip noise.

QSVR is highly vulnerable to adversarial attacks.

Abstract

Anomaly Detection (AD) is critical in data analysis, particularly within the domain of IT security. In this study, we explore the potential of Quantum Machine Learning for application to AD with special focus on the robustness to noise and adversarial attacks. We build upon previous work on Quantum Support Vector Regression (QSVR) for semisupervised AD by conducting a comprehensive benchmark on IBM quantum hardware using eleven datasets. Our results demonstrate that QSVR achieves strong classification performance and even outperforms the noiseless simulation on two of these datasets. Moreover, we investigate the influence of - in the NISQ-era inevitable - quantum noise on the performance of the QSVR. Our findings reveal that the model exhibits robustness to depolarizing, phase damping, phase flip, and bit flip noise, while amplitude damping and miscalibration noise prove to be more disruptive. Finally, we explore the domain of Quantum Adversarial Machine Learning by demonstrating that QSVR is highly vulnerable to adversarial attacks, with neither quantum noise nor adversarial training improving the model's robustness against such attacks.