Comparative Analysis of a Quantum SVM With an Optimized Kernel Versus Classical SVMs
Open Access
- 6 May 2025
- journal article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 82378-82387
Abstract
SupportVector Machine (SVM) is a widely used algorithm for classification, valued for its flexibility with kernels that effectively handle non-linear problems and high-dimensional data. Businesses across industries face challenges in improving customer retention and reducing churn, making predictive models essential for identifying at-risk customers and enhancing revenue. This paper investigates an optimized quantum embedding kernel for SVM (Quantum SVM - QSVM) applied to a public bank customer dataset, featuring variables on customer relationships and churn indicators. While focused on the financial sector, the methodology is broadly applicable for reducing churn and boosting revenue across industries. QSVM performance is compared to SVMs with established kernels, including Radial Basis Function (RBF), linear, polynomial, and sigmoid. Experiments varied the number of variables from two to seven to evaluate their impact on model performance and kernel behavior. The experiments were conducted on quantum simulators, which faced scalability challenges addressed using reduced datasets. Even so, this study sheds light on the potential of QSVMs to effectively manage increasing numbers of variables in predictive models, offering valuable insights into their capability to handle complex, high-dimensional data and their applicability in real-world scenarios.Keywords
Funding Information
- Brazilian Federal Agency
- Evaluation of Graduate Education
This publication has 29 references indexed in Scilit:
- Enhancing movie recommendations using quantum support vector machine (QSVM)The Journal of Supercomputing, 2024
- Exploring the Potential of Colloidal Quantum Dots for Near‐Infrared to Short‐Wavelength Infrared ApplicationsAdvanced Energy Materials, 2024
- On neural quantum support vector machinesQuantum Machine Intelligence, 2024
- Unraveling quantum computing system architectures: An extensive survey of cutting-edge paradigmsInformation and Software Technology, 2023
- A semi-agnostic ansatz with variable structure for variational quantum algorithmsQuantum Machine Intelligence, 2023
- Enhancing quantum support vector machines through variational kernel trainingQuantum Information Processing, 2023
- Kernel Functions of SVM: A Comparison and Optimal Solution2018
- Grover Meets Simon – Quantumly Attacking the FX-construction2017
- Early churn prediction with personalized targeting in mobile social gamesExpert Systems with Applications, 2017
- QSVM: A Support Vector Machine for Rule Extraction2015