Challenges
Datasets
Workspaces
Papers
Discussions
Leaderboard
Log inSign up
Challenges
Datasets
Workspaces
Papers
Discussions
Leaderboard
Blog
Job Board
Q3AS

© 2026 Aqora Quantum S.A.S.

TermsPrivacyLegal Notice

arXiv:1905.08220

Learning phase transitions: comparing PCA and SVM

Richard M. Woloshyn

Published May 20, 2019• 4 citations
arXivHTMLPDF

Abstract

A comparison of results from principal component analysis and support vector machine calculations is made for a variety of phase transitions in two-dimensional classical spin models.
DOI
10.48550/arxiv.1905.08220

Related papers

Exploration of phase behavior of classical spin systems using machine learning

Danilo Rodrigues de Assis Elias

•Mar 24, 2023•0 citations

A classification technique for condensed matter phases using a combination of PCA and SVM

Waleed K. Badawi, Z. M. Osman, Maha Sharkas +1 more

In quantum mechanical systems with exponentially large Hilbert space, the need to represent and identify states of quantum many-body system with few variables is of significant importance. The representation and identification of the states are based on the spin configurations of the ferromagnetic Ising model without knowledge of the respective Hamiltonian. The state identification process is of high importance in quantum technology applications and testing such as D-wave machine comparison to classical optimization algorithms using large number of qubits. This paper proposes a new method to classify phases and phase transitions in condensed matter systems, which can further be used in quantum technologies to identify the state of qubits. The proposed method is based on the combination of Principle component analysis (PCA) and support vector machine (SVM). The simulation results of the proposed method show that the trained model is able to identify the phase and phase transition with high accuracy in different Ising spin topologies with a variety of lattice sizes, while reducing the dimensionality of the feature space compared to existing optimization methods.

Photonics & Electromagnetics Research Symposium - Spring (PIERS-Spring)•May 1, 2017•6 citations

Construction of phase diagrams of spin models using unsupervised machine learning

Elguja Archemashvili

•Jan 1, 2025•0 citations

A Quantum Support Vector Machine-Based Approach for Customer Churn Prediction

S. Mohan Krishna P. Ratna Kumari

Customer churn prediction is important in the banking sector for improving customer retention and reducing revenue loss. Traditional Support Vector Machines (SVM) face scalability challenges when processing high-dimensional and non-linear customer data. This study proposes a Quantum Support Vector Machine (QSVM) for churn prediction by combining classical preprocessing with quantum machine learning. The data is cleaned, encoded, scaled, and reduced using Principal Component Analysis (PCA). The resulting features are encoded into quantum states, and a quantum kernel is used for classification. The model is implemented using Python and Qiskit and evaluated on the Qiskit Aer simulator. Performance is assessed using accuracy, precision, recall, and F1-score. The proposed QSVM achieves approximately 80% accuracy, demonstrating strong non-churn prediction and moderate churn identification. The results indicate the potential of quantum machine learning for banking churn prediction. Keywords : Customer Churn Prediction, Quantum Support Vector Machine (QSVM), Quantum Machine Learning, Banking, Principal Component Analysis (PCA), Quantum Kernel, Qiskit, Qiskit Aer, Machine Learning, Customer Retention

International Journal of Engineering Research and Science & Technology•Sep 24, 2026•0 citations

Detection of Phase Transitions in Quantum Spin Chains via Unsupervised Machine Learning

Yutaka Akagi, Nobuyuki Yoshioka, Hosho Katsura

Bulletin of the American Physical Society•Mar 5, 2019•0 citations

Comparative Performance Analysis of SVM and QSVM in Breast Cancer Diagnosis

Gita Kartika, Muhamad Akrom

The application of machine learning has shown significant progress in various fields, including clinical data classification. This study presents a comparative analysis between the classical Support Vector Machine (SVM) and Quantum Support Vector Machine (QSVM) for breast cancer classification using the Wisconsin Diagnostic Breast Cancer (WDBC) dataset. Principal Component Analysis (PCA) was applied as dimension reduction strategy across three configurations (PCA = 2, PCA = 4, and PCA = 6) to evaluate the sensitivity of both models to the feature space dimension. Preprocessing included log1p transformation on 22 skewed features, Manual SMOTE balancing, and MinMaxScaler normalization. SVM was optimized via GridSearchCV (80 parameter combinations, cv = 5), while QSVM employed a ZZFeatureMap quantum kernel (reps = 2, entanglement = full) on qasm_simulator. Experimental results demonstrate that SVM achieved superior accuracy across all configurations, with peak performance at PCA = 6 (Accuracy= 98.83%, AUC = 0.9985, Specificity = 1.0000, Sensitivity = 0.9688, Precision = 1.0000, NPV = 0.9817, and F1-score for malignant = 0.98). QSVM achieved its highest accuracy at PCA = 4 (92.98%), AUC = 0.9622, Sensitivity = 0.9219, Precision = 0.8939, NPV = 0.9524, and F1-score for malignant = 0.91. Across the three PCA configurations, QSVM exhibited a non-monotonic relationship between PCA dimensions and quantum kernel performance . These findings provide a quantitative benchmark for comparing quantum and classical machine learning in breast cancer diagnosis.

Journal of Applied Informatics and Computing•Oct 18, 2026•0 citations
In quantum mechanical systems with exponentially large Hilbert space, the need to represent and identify states of quantum many-body system with few variables is of significant importance. The representation and identification of the states are based on the spin configurations of the ferromagnetic Ising model without knowledge of the respective Hamiltonian. The state identification process is of high importance in quantum technology applications and testing such as D-wave machine comparison to classical optimization algorithms using large number of qubits. This paper proposes a new method to classify phases and phase transitions in condensed matter systems, which can further be used in quantum technologies to identify the state of qubits. The proposed method is based on the combination of Principle component analysis (PCA) and support vector machine (SVM). The simulation results of the proposed method show that the trained model is able to identify the phase and phase transition with high accuracy in different Ising spin topologies with a variety of lattice sizes, while reducing the dimensionality of the feature space compared to existing optimization methods.
Customer churn prediction is important in the banking sector for improving customer retention and reducing revenue loss. Traditional Support Vector Machines (SVM) face scalability challenges when processing high-dimensional and non-linear customer data. This study proposes a Quantum Support Vector Machine (QSVM) for churn prediction by combining classical preprocessing with quantum machine learning. The data is cleaned, encoded, scaled, and reduced using Principal Component Analysis (PCA). The resulting features are encoded into quantum states, and a quantum kernel is used for classification. The model is implemented using Python and Qiskit and evaluated on the Qiskit Aer simulator. Performance is assessed using accuracy, precision, recall, and F1-score. The proposed QSVM achieves approximately 80% accuracy, demonstrating strong non-churn prediction and moderate churn identification. The results indicate the potential of quantum machine learning for banking churn prediction. Keywords : Customer Churn Prediction, Quantum Support Vector Machine (QSVM), Quantum Machine Learning, Banking, Principal Component Analysis (PCA), Quantum Kernel, Qiskit, Qiskit Aer, Machine Learning, Customer Retention
The application of machine learning has shown significant progress in various fields, including clinical data classification. This study presents a comparative analysis between the classical Support Vector Machine (SVM) and Quantum Support Vector Machine (QSVM) for breast cancer classification using the Wisconsin Diagnostic Breast Cancer (WDBC) dataset. Principal Component Analysis (PCA) was applied as dimension reduction strategy across three configurations (PCA = 2, PCA = 4, and PCA = 6) to evaluate the sensitivity of both models to the feature space dimension. Preprocessing included log1p transformation on 22 skewed features, Manual SMOTE balancing, and MinMaxScaler normalization. SVM was optimized via GridSearchCV (80 parameter combinations, cv = 5), while QSVM employed a ZZFeatureMap quantum kernel (reps = 2, entanglement = full) on qasm_simulator. Experimental results demonstrate that SVM achieved superior accuracy across all configurations, with peak performance at PCA = 6 (Accuracy= 98.83%, AUC = 0.9985, Specificity = 1.0000, Sensitivity = 0.9688, Precision = 1.0000, NPV = 0.9817, and F1-score for malignant = 0.98). QSVM achieved its highest accuracy at PCA = 4 (92.98%), AUC = 0.9622, Sensitivity = 0.9219, Precision = 0.8939, NPV = 0.9524, and F1-score for malignant = 0.91. Across the three PCA configurations, QSVM exhibited a non-monotonic relationship between PCA dimensions and quantum kernel performance . These findings provide a quantitative benchmark for comparing quantum and classical machine learning in breast cancer diagnosis.