Classical machine learning methods have demonstrated strong performance in email phishing detection, but their ability to generalize against sophisticated phishing behaviors remains limited. Recent advances in quantum machine learning suggest that quantum kernel methods may capture complex feature interactions more effectively. This paper presents a batchwise ensemble evaluation framework that independently trains classical Support Vector Machine (SVM) and Quantum Support Vector Machine (QSVM) learning models using identical datasets. Due to quantum computational constraints, training is performed on fixed-size data batches and final predictions are obtained through majority voting across batches. Experimental results show that the QSVM ensemble reduces false negatives by 79 % compared to classical SVM, although at significantly higher computational cost, highlighting both the potential and current limitations of quantum kernel methods in the NISQ era.
Classical machine learning methods have demonstrated strong performance in email phishing detection, but their ability to generalize against sophisticated phishing behaviors remains limited. Recent advances in quantum machine learning suggest that quantum kernel methods may capture complex feature interactions more effectively. This paper presents a batchwise ensemble evaluation framework that independently trains classical Support Vector Machine (SVM) and Quantum Support Vector Machine (QSVM) learning models using identical datasets. Due to quantum computational constraints, training is performed on fixed-size data batches and final predictions are obtained through majority voting across batches. Experimental results show that the QSVM ensemble reduces false negatives by 79 % compared to classical SVM, although at significantly higher computational cost, highlighting both the potential and current limitations of quantum kernel methods in the NISQ era.
Quantum Machine Learning (QML) provides a pathway for analysing nonlinear relationships through quantum- enhanced feature spaces. Although quantum models are theoreti- cally well suited for high-complexity data, current hardware and simulator limitations restrict their use to small datasets. For this reason, the original large Web Phishing Detection Dataset was reduced to 200 representative samples through dimensionality reduction, allowing controlled experimentation and enabling evaluation of quantum behaviour with manageable computational cost. QML research commonly follows this approach, testing algorithms on small datasets first, because quantum feature maps can generate exponentially expressive Hilbert-space embeddings even from limited inputs. In this study, multiple Quantum Support Vector Machine (QSVM) models were implemented using fidelity-based quantum kernels and Pauli feature maps. Models with 2 to 10 qubits were evaluated to observe how circuit width affects performance. The two-qubit QSVM achieved the best quantum result with 72 % accuracy, whereas deeper circuits showed declining performance, indicating overfitting and reduced generalization on limited data. In contrast, the classical SVM attained 97 % accuracy with significantly lower computational effort. These findings highlight the challenges and potential of QML for cybersecurity tasks. Future work will explore circuit- expressivity tuning, optimized quantum kernels, hybrid quantum- classical approaches, and scalability to larger phishing datasets as quantum hardware evolves.
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.
Cyberbullying has become much more common due to the proliferation of social media, which has devastating effects on users' mental health and relationships. Automatic identification of abusive textual content has therefore become an important research problem. Cyberbullying detection has seen extensive use of classical machine learning approaches, notably Support Vector Machines (SVMs); however, their effectiveness can be limited when handling high-dimensional and context-rich textual features. Recent developments in quantum computing introduce new computational paradigms that may enable more expressive feature representations through quantum-enhanced learning models. In this study, we explore the potential of a hybrid. Nowadays classical-quantum architecture to apply a Quantum Support Vector Machine (QSVM) to detect cyberbullying. Experiments were performed on a publicly available Kaggle dataset containing 35,824 labeled tweets. Before TF-IDF feature extraction, the text data underwent preprocessing via cleaning and tokenization. Dimension reduction was carried out using Principal Component Analysis (PCA) in respect of qubit constraints. Then, they were embedded into quantum states using angle embedding. The accuracy, precision, recall, and F1-score measures were used to compare QSVM models with various qubit configurations (2, 4, and 6), as well as a traditional SVM baseline, under an 80-20 stratified train-test split. The findings indicate that the quality of QSVM performance is highest with the increase in qubit numbers and the maximum accuracy of the quantum computing method is 60 percent according to six qubits, whereas the classical SVM achieves 67.5 percent. Although the quantum model currently incurs higher computational overhead due to simulator-based implementation, increasing qubit dimensionality enhances its ability to capture complex and non-linear linguistic patterns. These observations indicate that QSVM-based cyberbullying detection remains exploratory at present, with scope for further improvement as quantum hardware and algorithms advance.
Quantum computing holds significant potential to accelerate machine learning algorithms, especially in solving optimization problems like those encountered in Support Vector Machine (SVM) training. However, current QUBO-based Quantum SVM (QSVM) methods rely solely on binary optimal solutions, limiting their ability to identify fuzzy boundaries in data. Additionally, the limited qubit count in contemporary quantum devices constrains training on larger datasets. In this paper, we propose a probabilistic quantum SVM training framework suitable for Coherent Ising Machines (CIMs). By formulating the SVM training problem as a QUBO model, we leverage CIMs' energy minimization capabilities and introduce a Boltzmann distribution-based probabilistic approach to better approximate optimal SVM solutions, enhancing robustness. To address qubit limitations, we employ batch processing and multi-batch ensemble strategies, enabling small-scale quantum devices to train SVMs on larger datasets and support multi-class classification tasks via a one-vs-one approach. Our method is validated through simulations and real-machine experiments on binary and multi-class datasets. On the banknote binary classification dataset, our CIM-based QSVM, utilizing an energy-based probabilistic approach, achieved up to 20% higher accuracy compared to the original QSVM, while training up to 104 times faster than simulated annealing methods. Compared with classical SVM, our approach either matched or reduced training time. On the IRIS three-class dataset, our improved QSVM outperformed existing QSVM models in all key metrics. As quantum technology advances, increased qubit counts are expected to further enhance QSVM performance relative to classical SVM.
Abstract Support vector machines (SVMs) are widely used for classification, but become computationally expensive as datasets grow and data become high-dimensional or nonlinearly separable. Quantum support vector machines (QSVMs) enhance the traditional SVMs by replacing conventional feature mappings with quantum-implemented embeddings that project data into high-dimensional Hilbert spaces, potentially improving expressive power and efficiency. However, training QSVMs on current noisy intermediate-scale quantum (NISQ) devices remains challenging, as constructing the quantum kernel matrix requires repeated circuit executions and scales quadratically with dataset size. We propose NQ-SVM , a hybrid quantum-classical method that incorporates the Nystr\"om approximation to enhance the scalability of QSVMs. By selecting a small set of landmark points, the Nystr\"om method yields a low-rank kernel approximation, reducing complexity from (O(n2)) to (O(nm)), while preserving the quantum advantages in handling high-dimensional, non-linear data. Extensive simulations demonstrate that NQ-SVM achieves comparable classification accuracy to classical SVMs and full QSVMs, while substantially reducing training time and kernel evaluation costs. These results underscore the potential of NQ-SVM as a scalable and practical hybrid framework for near-term quantum machine learning (QML).
This article explores the potential of Quantum Machine Learning (QML), specifically assessing a Quantum Support Vector Machine (QSVM) and a Variational Quantum Classifier (VQC) for detecting anomalies in real-world financial transaction data. While these QML methods outperform statistical methods, they fall short of cutting-edge deep learning techniques. To bridge this gap, we propose a hybrid quantum-classical ensemble framework that leverages the strengths of both domains. We demonstrate its effectiveness in detecting phishing in Ethereum transaction networks by combining complementary algorithms. The QSVM, whether used individually or in an ensemble, consistently delivered the lowest false negatives and higher recall rates, that are crucial for anomaly detection. To enhance individual models, we encoded the data using novel cascaded Quantum Random Access Coding (QRAC) schemes and compared it with the popular encoding ZZ feature map on both simulators and the IBM Heron quantum processor. For both QSVM and VQC, we consistently observed improvements (13% for QRAC-VQC and 3% for QRAC-QSVM) of QRAC over the ZZ feature map. Notably, certain QML algorithms exhibit remarkable resilience on the IBM Heron quantum processor, approaching simulator-level performance on devices with high quantum volume. This observation underscores the promise of QML despite hardware limitations.