Evaluating the Potential of Quantum Machine Learning in Cybersecurity: A Case-Study on PCA-based Intrusion Detection Systems

TL;DR

The paper assesses the potential impact of fault-tolerant quantum machine learning on cybersecurity, focusing on PCA-based intrusion detection as a case study. It develops an evaluation framework that compares QML with classical ML while accounting for data-loading costs, QRAM memory access, and dataset-dependent parameters, using simulated quantum subroutines for model extraction. The findings indicate that quantum advantage in query complexity arises only for very large datasets and substantial hardware resources, with practical gains hampered by data-loading bottlenecks and QRAM latency; thus near-term benefits in IDS are limited. The framework offers a structured, transferable approach for practitioners to forecast QML benefits in cybersecurity as quantum hardware matures.

Abstract

Quantum computing promises to revolutionize our understanding of the limits of computation, and its implications in cryptography have long been evident. Today, cryptographers are actively devising post-quantum solutions to counter the threats posed by quantum-enabled adversaries. Meanwhile, quantum scientists are innovating quantum protocols to empower defenders. However, the broader impact of quantum computing and quantum machine learning (QML) on other cybersecurity domains still needs to be explored. In this work, we investigate the potential impact of QML on cybersecurity applications of traditional ML. First, we explore the potential advantages of quantum computing in machine learning problems specifically related to cybersecurity. Then, we describe a methodology to quantify the future impact of fault-tolerant QML algorithms on real-world problems. As a case study, we apply our approach to standard methods and datasets in network intrusion detection, one of the most studied applications of machine learning in cybersecurity. Our results provide insight into the conditions for obtaining a quantum advantage and the need for future quantum hardware and software advancements.

Figures (8)

• Figure 1: Framework for evaluating speedups with quantum algorithms.
• Figure 2: Detection procedures of the three considered PCA-based anomaly detection classifiers.
• Figure 3: Quantum PCA model extraction.
• Figure 4: Running time comparison of classical (green) and quantum (blue) algorithms. In Plot \ref{['fig:pcc_runtime_comparisonKDD']} we show the comparison of PCA and QPCA of Section \ref{['ssec:pcckdd']} (KDDCUP99). In Plot \ref{['fig:pcc_cicids_runtime']} we show the running times of the algorithms discussed in Section \ref{['sec:pcc_mm_cicids']} (CICIDS2017). In Plot \ref{['fig:runtimePCAloss_']} we show the comparison between PCA and QPCA over the CICIDS2017 dataset, as per Section \ref{['ssec:pca_loss']}.
• Figure 5: Running time comparison of classical (green) and quantum (blue) algorithms for Rec. Loss over DARKNET.

Theorems & Definitions (10)

• Definition B.1: Quantum access to a matrix kerenidis2020quantumGRADIENT
• Theorem B.2: Implementing quantum operators using an efficient data structure kerenidis2016quantumREC
• Definition B.3: Memory parameter $\mu(X)$ kerenidis2020quantumGRADIENT
• Theorem B.4: Tomographykerenidis2020quantumIP
• Theorem B.5: Phase estimation nielsen2010quantum
• Theorem B.6: Amplitude estimationbrassard2002quantum
• Theorem B.7
• Theorem B.8: Quantum binary search for the singular value threshold qadra
• Theorem B.9: Top-k singular vectors extraction qadra
• Theorem B.10: $q$-means qmeans