The conceptual framework of quantum computing has recently emerged as promising to enhance machine learning in complex non-stationary settings of sensing, but its use is still challenged by hardware noise and concept drift in streaming data. This paper demonstrates a noise-aware quantum kernel method of online anomaly detection on high-velocity sensor streams in changing data distributions. To represent the signal magnitude and dynamics, a two-dimensional temporal embedding is used and quantum kernel evaluations are done via Monte Carlo sampling of explicitly depolarizing and readout noisy models to simulate near-term hardware constraints. The resultant uncertainty is harnessed in confidence-weighted anomaly scoring and adaptive thresholding which allows effective drift-tolerant detection with no labeled data. The suggested approach is tested on streaming sensor benchmarks using injected concept drift and evaluated against radial basis function (RBF) kernels and ideal (noise-free) quantum kernels. Through experimental results, it is shown that although classical RBF has the highest ROC-AUC of 0.9762 and PR-AUC of 0.4444, the proposed quantum noise-aware RBF generates impressive results achieve ROC-AUC of 0.9017 and PR-AUC of 0.4134, which means most of the lost performance is associated with quantum noise. The findings emphasize the significance of the explicit modeling of quantum uncertainty and show the practical feasibility of quantum machine learning in the real-time non-stationary deployment-oriented anomaly detection in sensor systems.
The conceptual framework of quantum computing has recently emerged as promising to enhance machine learning in complex non-stationary settings of sensing, but its use is still challenged by hardware noise and concept drift in streaming data. This paper demonstrates a noise-aware quantum kernel method of online anomaly detection on high-velocity sensor streams in changing data distributions. To represent the signal magnitude and dynamics, a two-dimensional temporal embedding is used and quantum kernel evaluations are done via Monte Carlo sampling of explicitly depolarizing and readout noisy models to simulate near-term hardware constraints. The resultant uncertainty is harnessed in confidence-weighted anomaly scoring and adaptive thresholding which allows effective drift-tolerant detection with no labeled data. The suggested approach is tested on streaming sensor benchmarks using injected concept drift and evaluated against radial basis function (RBF) kernels and ideal (noise-free) quantum kernels. Through experimental results, it is shown that although classical RBF has the highest ROC-AUC of 0.9762 and PR-AUC of 0.4444, the proposed quantum noise-aware RBF generates impressive results achieve ROC-AUC of 0.9017 and PR-AUC of 0.4134, which means most of the lost performance is associated with quantum noise. The findings emphasize the significance of the explicit modeling of quantum uncertainty and show the practical feasibility of quantum machine learning in the real-time non-stationary deployment-oriented anomaly detection in sensor systems.
Quantum Kernel Methods for Anomaly Detection in High-Velocity Data Streams introduces a novel framework leveraging quantum computing principles to address critical challenges in real-time anomaly detection. By combining the expressive power of quantum-enhanced feature spaces with classical machine learning techniques, the work presents a hybrid architecture capable of identifying subtle anomalies in complex, high-dimensional streaming data. The framework incorporates specialized quantum feature maps that efficiently encode temporal and distributional properties of data streams, while adaptation mechanisms respond to concept drift and evolving patterns. Through systematic experimental evaluation across synthetic and real-world datasets from financial transactions, network security, and industrial systems, the approach demonstrates superior detection performance particularly for complex nonlinear patterns in high-dimensional spaces. The quantum-classical implementation addresses current hardware constraints through optimization techniques and targeted resource allocation, establishing specific conditions where quantum advantage emerges for operational anomaly detection scenarios.
Unsupervised concept drift detection in non-stationary tabular streams is challenging because real systems rarely provide labelled annotations. This work presents a Hybrid Quantum-Classical Concept Drift Detection Framework that integrates a classical RBF-kernel Maximum Mean Discrepancy (MMD) detector with a 4-qubit quantum fidelity-kernel MMD computed using PennyLane. Both detectors operate on sliding reference-test windows and perform permutation-based significance testing to quantify distributional change. Their outputs are combined through a variance-gated hybrid fusion mechanism that leverages quantum expressivity while stabilizing noisy quantum responses. Since real drift labels are unavailable, controlled synthetic mean-shift drifts are injected into selected windows, enabling a lightweight meta-learner to evaluate detection quality without altering the unsupervised nature of the core framework. Experiments on three large financial datasets show that the hybrid detector consistently outperforms standalone classical and quantum MMD baselines, achieving up to 10/15 injecteddrift detections under adaptive thresholds, while the meta-learner achieves accuracy above 97% and ROC-AUC above 0.98. These results demonstrate that combining classical and quantum kernel statistics with reliability-aware fusion offers a practical and effective approach for drift detection in high-dimensional financial data streams.
Anomaly detection, an important task in data analysis, involves identifying observations or events that deviate in some way from the rest of the data. Machine learning techniques have shown success in automating this process by detecting hidden patterns and deviations in large-scale data. The potential of quantum computing for machine learning has been widely recognized, leading to extensive research efforts to develop suitable quantum machine learning (QML) algorithms. In particular, the search for QML algorithms for near-term NISQ devices is in full swing. However, NISQ devices pose additional challenges due to their limited qubit coherence times, low number of qubits, and high error rates. Kernel methods based on quantum kernel estimation have emerged as a promising approach to QML on NISQ devices, offering theoretical guarantees, versatility, and compatibility with NISQ constraints. Especially support vector machines (SVM) utilizing quantum kernel estimation have shown success in various supervised learning tasks. However, in the context of anomaly detection, semisupervised learning is of great relevance, and yet there is limited research published in this area. This paper introduces an approach to semisupervised anomaly detection based on the reconstruction loss of a support vector regression (SVR) with quantum kernel. This novel model is an alternative to the variational quantum and quantum kernel one-class classifiers, and is compared to a quantum autoencoder as quantum baseline and a SVR with radial-basis-function (RBF) kernel as well as a classical autoencoder as classical baselines. The models are benchmarked extensively on 10 real-world anomaly detection data sets and one toy data set and it is shown that our SVR model with quantum kernel performs better than the SVR with RBF kernel as well as all other models, achieving highest mean AUC over all data sets. In addition, our QSVR outperforms the Quantum autoencoder on 9 out of 11 data sets.
The detection of android malware by machine learning techniques continues to face challenges with the increasing diversity of malware. The classical machine learning approach based on Support Vector Machines (SVM) with radial basis function (RBF) kernel has been shown to perform reliably on static features such as permissions and API calls. Meanwhile, an alternative quantum machine learning approach based on Quantum Support Vector Machines (QSVM) has been introduced by replacing classical similarity measures with quantum kernels. However, the performance of QSVM under real-world noisy quantum environments remains unknown. In this work, we investigate the performance of QSVM for android malware detection by employing a stratified 150-sample subset of the TUANDROMD dataset. We reduce the feature space from 241 dimensions to four principal components according to current qubit limitations while retaining 59.4% of the total variance of the feature space. We compare the performance of QSVM with a classical machine learning approach based on RBF-SVM by simulating depolarizing gate errors and measurement noise via the Qiskit Aer package. The experimental results show that the classical SVM has 90.4% accuracy with a 5.6% false negative rate, while QSVM has only 83.7% under ideal simulation and gradually reduces with the increase in noise. The statistical analysis has proved that the gap between the two is statistically significant. In summary, although QSVM has shown its feasibility, it has failed to compete with the classical algorithms under current hardware constraints.
Quantum computing offers potential advancements in machine learning. However, the impact of concept drift remains unexplored when introducing data streams into quantum machine learning models. Hence, this study is the first to examine the impact of concept drift on quantum support vector machines and quantum neural networks, contrasting them with their classical counterparts. Our study is based on two data streams provided by the RiverML Library, namely, the real KDDCUP99 dataset and the synthetic RandomRBFDrift dataset. We demonstrate that the quantum models under study are more susceptible to concept drift compared with their classical counterparts in the KDDCUP99 dataset. This is seen in the results as more severe drops with longer recovery duration exist in the prequential area under the curve (AUC) of the quantum models' receiver operating characteristics curve (ROC). On the other hand, comparable results are observed in the RandomRBFDrift dataset. As a mitigation strategy for concept drift in quantum models, we propose an optimal encoding strategy based on a genetic optimization that maximizes the AUC despite concept drift. The proposed strategy determines, based on the dataset, an optimal encoding of the classical data into quantum states using a combination of angle, amplitude, and basis encoding, optimal repetitions, circuit depth, etc. Our results demonstrate that the proposed strategy enhances the performance of the quantum models in the presence of concept drift, showcasing up to 5% improvement in the model accuracy. This study provides valuable insights into the effects and mitigation of concept drift in quantum machine learning and opens the door for future research.
Anomaly detection (AD) underpins fraud prevention, network security, industrial condition monitoring, medical diagnosis, and searches for new physics. However, it remains challenging due to rare anomalies, scarce labeled data, and the high dimensionality and strong correlation of sensor data. Many classical machine learning (ML) algorithms are designed to address AD in various settings, but the growth of quantum computing opens a new path for these problems. Quantum machine learning (QML) offers new tools, from exploring exponentially large Hilbert spaces and promising speed-ups to the native processing of quantum data. Thus, this review connects the two fields. We first present the basics of quantum computing, quantum algorithms, and QML with the most influential literature, and formalize the classical anomaly scores that quantum methods inherit. We then systematically review 30 primary studies of quantum anomaly detection (QAD) published between 2018 and 2026, express each method family through its anomaly score, and organize the field into a taxonomy of six families: quantum kernel methods, quantum autoencoders, quantum generative models, variational quantum classifiers, quantum linear-algebra methods, and detectors for quantum data. A cross-study analysis shows that the field has moved from asymptotic speed-up proposals to small hybrid experiments, that kernels and autoencoders dominate, that only nine studies run on gate-based hardware, and that none yet demonstrates a robust, practically relevant advantage over well-tuned classical detectors. We close with the main obstacles, such as data embedding, barren plateaus, kernel concentration, and weak baselines, and a research roadmap that prioritizes quantum data, fair benchmarks against classical baselines, and robust detectors for sequential and distributed data.