ROBUSTNESS EVALUATION OF QUANTUM MACHINE LEARNING FOR NETWORK SECURITY: A SYSTEMATIC MAP AND CRITICAL APPRAISAL

Summary

This systematic map examines whether quantum machine learning network attack detectors have been tested for robustness, screening Scopus and IEEE Xplore records from 2017 onward. Of 33 selected studies, robustness was mostly tested as simulated quantum noise (17 studies) or distribution shift within single benchmarks (10), with attacks on the detector tested in nine studies. The authors conclude that the claim that QML makes network detectors more robust is not currently supported by the evidence.