Reliable anomaly detection in subsea infrastructure is critical, as overlooked faults may result in severe environmental damage and substantial financial losses. Although classical machine learning approaches provide effective solutions, they often require significant computational resources and may struggle to capture the complex, high-dimensional correlations present in multivariate sensor data. In this work, we present a comprehensive evaluation of quantum and hybrid quantum–classical architectures using the 3W dataset, a large-scale industrial benchmark comprising 2,184 real and simulated subsea events. We develop an end-to-end pipeline that transforms raw multivariate time-series signals into quantum-compatible feature representations. Four hybrid architectures are systematically compared against six classical baselines in both supervised and unsupervised settings. Our results demonstrate that a compact hybrid model (Hybrid-A) achieves an F1-score of 0.9737 using only 177 trainable parameters, matching the performance of a 10,000-parameter Random Forest and exhibiting substantially higher parameter efficiency than conventional neural networks. In the unsupervised scenario, the proposed Quantum Autoencoder outperforms its classical counterpart, improving F1-score by 5.35% and ROC-AUC by 15.28% while reducing parameter count by 42%. These findings indicate that quantum-enhanced models can achieve competitive performance with significantly reduced model complexity, highlighting their potential for deployment in resource-constrained, safety-critical edge environments. Keywords: Quantum Machine Learning, Industrial Anomaly Detection, Hybrid Quantum-Classical Models, 3W Dataset, Edge Deployment, Parameter Efficiency
Reliable anomaly detection in subsea infrastructure is critical, as overlooked faults may result in severe environmental damage and substantial financial losses. Although classical machine learning approaches provide effective solutions, they often require significant computational resources and may struggle to capture the complex, high-dimensional correlations present in multivariate sensor data. In this work, we present a comprehensive evaluation of quantum and hybrid quantum–classical architectures using the 3W dataset, a large-scale industrial benchmark comprising 2,184 real and simulated subsea events. We develop an end-to-end pipeline that transforms raw multivariate time-series signals into quantum-compatible feature representations. Four hybrid architectures are systematically compared against six classical baselines in both supervised and unsupervised settings. Our results demonstrate that a compact hybrid model (Hybrid-A) achieves an F1-score of 0.9737 using only 177 trainable parameters, matching the performance of a 10,000-parameter Random Forest and exhibiting substantially higher parameter efficiency than conventional neural networks. In the unsupervised scenario, the proposed Quantum Autoencoder outperforms its classical counterpart, improving F1-score by 5.35% and ROC-AUC by 15.28% while reducing parameter count by 42%. These findings indicate that quantum-enhanced models can achieve competitive performance with significantly reduced model complexity, highlighting their potential for deployment in resource-constrained, safety-critical edge environments. Keywords: Quantum Machine Learning, Industrial Anomaly Detection, Hybrid Quantum-Classical Models, 3W Dataset, Edge Deployment, Parameter Efficiency
Reliable anomaly detection in subsea infrastructure is critical: overlooked faults can lead to catastrophic environmental damage and billions in losses. While classical machine learning offers a partial solution, it often demands substantial computational resources and struggles with the complex, high-dimensional correlations inherent in sensor data. This study presents the first comprehensive evaluation of quantum and hybrid quantum-classical architectures on the 3W dataset, a massive industrial benchmark of 2,184 real and simulated subsea events. We develop an end-to-end pipeline that transforms raw multivariate time series into quantum-ready features, comparing four hybrid architectures against six classical baselines. Our results are compelling: a simple hybrid model (Hybrid-A) achieves an F1 score of 0.9737 with just 177 parameters-matching the performance of a 10,000-parameter Random Forest while being 56× more parameter-efficient than standard neural networks. In the unsupervised setting, our Quantum Autoencoder significantly outperforms its classical counterpart, improving F1 by 5.35% and ROC-AUC by 15.28% with 42% fewer parameters. These findings demonstrate that quantum-enhanced models can deliver state-of-the-art performance with a fraction of the complexity, making them ideal for resource-constrained edge deployment in safety-critical systems.
Hyperspectral anomaly detection (HAD) is an important unsupervised question within a field called remote sensing where the problem seeks to be the recognition of the presence of spectrally anomalous and previously unobserved targets within complex background terrains. To address these shortcomings, the current publication presents a new method for Hyperspectral Anomaly Detection (HQML-HAD) to address them, i.e. a Hybrid Quantum Machine Learning framework to synergies the classical spectral-spatial representation learning with a quantum enhanced anomaly decision modelling. The proposed architecture uses a lightweight classical encoder to realize spectral dimensionality reduction together with context estimation in space which guarantees efficiency of the preprocessing while respecting the qubit resource constraints. The evaluation of the framework is done on benchmark real-world HAD data sets, which contain highly unbalanced scenes with sparse distributions of anomalous targets under heterogeneous background scenes. All of the experiments were run in Google Cirq situated quantum simulation environment deployed on Google Collab infrastructure thus allowing scalable hybrid quantum classical experimentation without requiring access to physical quantum hardware. Experimental results show that the proposed HQML- HAD framework achieves an overall detection accuracy of 67.82%, Area Under the ROC Curve (AUC)=0.712, false alarm rate=14.6%, F1-score=0.684, and outperforms the performance of conventional statistical and autoencoder based baselines in the same experimental conditions. These results validate the effectiveness and the practicality of hybrid quantum machine learning approach to the hyperspectral anomaly detection for realistic remote sensing applications.
There is a pressing need for anomaly detection algorithms that can handle large amounts of diverse traffic data due to the quick growth of networked systems and the sophistication of contemporary cyberattacks. Despite the widespread use of conventional machine learning techniques for network intrusion detection, feature representation constraints, dataset imbalance, and the computational expense of scaling to vast problem spaces sometimes restrict their efficacy. In the meantime, new developments in quantum-assisted computation have opened up exciting possibilities for using quantum state changes to improve the feature learning process. Using the NF-UNSWNB15-v2 benchmark, this research presents HybridQCNN, a unique quantum-classical computing framework for binary classification of network traffic.The proposed architecture combines standard preprocessing techniques with an 8-qubit Parameterized Quantum Circuit (PQC) featuring multiple variational layers, where IsingXX and IsingYY entangling operations are applied to encode interfeature correlations, and a subsequent fully connected classical layer handles the final binary decision. A rigorous leakage-free data partitioning protocol was enforced, ensuring that all preprocessing transformations—including feature encoding, normalization, and threshold computation—were derived exclusively from the training subset, thereby preventing any information leakage from held-out partitions during evaluation. Experimental validation on a class-balanced subset of 60,000 traffic records yielded a classification accuracy of 97.33%, an F1-score of 0.9730, and a malicious traffic detection precision of 98.38%. Furthermore, the model achieved an AUC-ROC of 0.9861, reflecting strong class-level discriminability. Despite this performance, the entire model contains only 489 trainable parameters, underscoring the parameter efficiency of hybrid architectures relative to conventional deep learning baselines. These results suggest that integrating variational quantum circuits with classical inference modules provides an effective and scalable solution for network anomaly detection.
This research explores the integration of quantum computing and machine learning for anomaly detection in critical infrastructure sensor networks. Anomaly detection is crucial for maintaining the reliability and security of these networks, where early detection of faults or intrusions can prevent catastrophic failures. Traditional machine learning methods often struggle with the high-dimensional and complex data generated by these systems. Hybrid quantum approaches, combining quantum algorithms with classical machine learning models, offer a promising solution by leveraging quantum computational advantages, such as quantum feature mapping and kernel methods. This paper investigates how these hybrid models can improve detection accuracy, reduce processing time, and handle large-scale data more efficiently. The potential of quantum machine learning for anomaly detection in critical infrastructure is discussed, along with the challenges posed by current quantum hardware limitations and future directions for research.
Extending our prior work [1] which introduced a hybrid Quantum Neural Network (QNN) and Quantum Autoencoder (QAE) pipeline for univariate temperature forecasting and anomaly detection on edge IoT devices, this paper advances the framework to multivariate anomaly detection across correlated pressure and temperature sensors. While [1] demonstrated the feasibility of quantum-enhanced edge analytics, its QAE was limited to single-variable reconstruction, failing to detect anomalies that manifest only as joint deviations across multiple sensors. In this study, we retain the identical QNN+QAE architecture including 8-qubit AngleEmbedding and 3-layer StronglyEntanglingLayers circuits, and PCA-based dimensionality reduction while introducing two key extensions: (1) retargeting the QNN to forecast pressure instead of temperature, and (2) crucially, training the QAE on multivariate input vectors comprising both pressure and temperature data, enabling joint anomaly scoring via combined reconstruction error. Evaluated on real-world IoT data deployed on a Raspberry Pi 5, the QNN achieves pressure forecasting accuracy comparable to prior temperature results(R2=0.91,MAE=0.0499,RMSE=0.0636). More significantly, the multivariate QAE achieves an anomaly detection F1 score of 0.92, substantially outperforming univariate QAEs trained separately on pressure(0.83)and temperature(0.85), particularly in detecting slow correlated drifts and sensor decalibration scenarios that single-variable methods miss. The multivariate QAE adds no additional trainable parameters and incurs only14%inference latency overhead compared to univariate counterparts, preserving edge deployment feasibility. To our knowledge this work represents one of the early demonstrations of a hybrid quantum-classical framework for multivariate anomaly detection on edge IoT hardware, demonstrating the feasibility of multivariate hybrid quantum-classical analytics without additional parameter or architectural overhead, directly building upon and extending the univariate capabilities reported in [1]
This work investigates variational quantum autoencoders (QAE) for one-class anomaly detection under strict parametric constraints, shifting the evaluation focus from absolute performance to performance–capacity tradeoffs. We implement a pure QAE baseline and a Hybrid QAE with classical compression/decoding around a quantum latent block, comparing both against classical baselines - AE, VAE, Isolation Forest, and One-Class SVM - on the NSL-KDD and ECG5000 benchmarks. The central contribution is a fairnessoriented evaluation protocol combining standard detection metrics (AUC-ROC, AUC-PR, and F1) with a performance-density measure (AUC-ROC per 1,000 trainable parameters) and a fair-budget comparison against a compact classical AE. On NSL-KDD, the QAE baseline achieves 0.9612 AUC-ROC with only 476 parameters versus 0.9679 for a classical AE with 5,580 parameters, while attaining 2.0193 AUC-ROC/1k parameters against 0.1735 for the classical AE - an order-of-magnitude improvement in parametric efficiency. Under a fairbudget setting, the QAE baseline also surpasses the compact classical AE on NSL-KDD (0.9612 vs. 0.9405). In contrast, ECG5000 favors classical methods, indicating domain dependence rather than universal quantum advantage. Overall, quantum and hybrid autoencoders are not universally superior, but deliver competitive anomaly detection with remarkably high parametric efficiency.