Performance Analysis of Quantum Support Vector Classifiers and Quantum Neural Networks
Tomás Villalba-Ferreiro, Eduardo Mosqueira-Rey, Diego Alvarez-Estevez
TL;DR
The paper assesses whether quantum classifiers (QSVC) and quantum neural networks (QNN) outperform classical counterparts as problem difficulty increases, using Iris and MNIST-PCA datasets. It implements QSVCs with regular and trainable quantum kernels and models QNNs as variational quantum classifiers, while exploring feature maps, ansatz configurations, losses, optimizers, and comparing PennyLane with Qiskit. Results show QSVCs are generally more consistent, while QNNs gain relative advantage on higher-complexity data; increasing the quantum component improves performance and hyperparameter tuning strongly influences accuracy. The work highlights the potential of quantum machine learning for complex classification, providing guidance on model selection and optimization and identifying Qiskit as a favorable framework for quantum kernel tasks.
Abstract
This study explores the performance of Quantum Support Vector Classifiers (QSVCs) and Quantum Neural Networks (QNNs) in comparison to classical models for machine learning tasks. By evaluating these models on the Iris and MNIST-PCA datasets, we find that quantum models tend to outperform classical approaches as the problem complexity increases. While QSVCs generally provide more consistent results, QNNs exhibit superior performance in higher-complexity tasks due to their increased quantum load. Additionally, we analyze the impact of hyperparameter tuning, showing that feature maps and ansatz configurations significantly influence model accuracy. We also compare the PennyLane and Qiskit frameworks, concluding that Qiskit provides better optimization and efficiency for our implementation. These findings highlight the potential of Quantum Machine Learning (QML) for complex classification problems and provide insights into model selection and optimization strategies