Exploring Quantum Machine Learning for Electroencephalogram Classification

• Published in: IEEE Symposium on Computer Applications and Industrial Electronics (Online) pp. 392 - 397
• Main Authors: Ho, Raymond, Hung, Kevin
• Format: Conference Proceeding
• Language: English
• Published: IEEE 20.05.2023
• Subjects: dementia classification; EEG classification; Electroencephalography; Machine learning; Performance evaluation; Quantum algorithm; Quantum entanglement; quantum machine learning; Qubit; Shape
• ISSN: 2836-4317
• DOI: 10.1109/ISCAIE57739.2023.10165407

Quantum machine learning (QML) is a relatively new discipline emerging from the concepts of machine learning and quantum computing, whereby quantum algorithms are used to solve machine learning tasks. This paper explores the use of quantum machine learning for electroencephalogram (EEG) classification. In particular, a previously proposed EEG feature extraction method and classification framework for classifying dementia subjects were followed in this study. A quantum classifier replaced the classical classifier component of the framework, and the classification accuracies between the quantum and classical classifiers were compared. This study has demonstrated that applying QML in healthy-dementia classification can be implemented using near-term quantum devices or quantum simulators with moderate performance. The quantum classifier achieved an overall classification accuracy of 81.67% and 79.17% in a train-test split performance test and an n \times k-fold cross-validation test, respectively. However, the quantum approach did not produce higher classification accuracies than the classical classifier. Despite the promise of quantum advantages, further investigation and optimization are required to improve its effectiveness.