Hybrid quantum–classical frameworks for anomaly detection

Quantum anomaly detection is an emerging research area at the intersection of quantum computing and machine learning, aiming to identify irregularities or deviations in data by utilising quantum-enhanced models. The proposed study seeks to develop scalable, hybrid quantum–classical frameworks for anomaly detection that are optimized for current noisy intermediate-scale quantum (NISQ) hardware. Although previous work has shown that quantum algorithms such as quantum kernel principal component analysis, quantum autoencoders, and quantum one-class classifiers can detect anomalies effectively, most of these demonstrations remain confined to idealized, small-scale simulations. The research will focus on addressing this limitation by designing algorithms that can process larger datasets while remaining robust to quantum noise. The study will integrate classical preprocessing for dimensionality reduction with quantum models that perform anomaly scoring on compressed representations, thereby reducing data-loading overheads. It will also explore parameter-efficient variational circuits that maintain detection accuracy under realistic noise conditions, making them suitable for implementation on currently available quantum processors. Additionally, the project will emphasize interpretability by developing techniques to explain the decisions made by quantum anomaly detectors, enhancing their usability in critical domains such as finance, cybersecurity, and network monitoring. The expected outcomes include a validated, hardware-compatible quantum anomaly detection framework, new insights into the behavior of hybrid quantum–classical models under noisy conditions, and practical applications demonstrating quantum advantage in detecting complex, high-dimensional anomalies. This work builds upon the ongoing efforts at Deakin University and IIT Madras, extending their contributions in quantum deep learning and cybersecurity toward a robust, deployable architecture for next-generation anomaly detection systems. To validate the approach, the project will define standardized benchmarks, curate representative datasets, and report metrics such as precision–recall curves, false-positive rates, and inference latency. Prototypes will be implemented using widely adopted quantum software frameworks and executed across multiple hardware backends to assess portability. Reproducibility will be prioritized through versioned code and experiment logs. Stakeholder feedback from domain partners will inform iterative refinements, risk assessment, and pathways to pilot deployments.