Publicação
Classification and clustering using swap test as distance metric
| Resumo: | This master’s thesis explores the advantages of using a quantum-based distance metric in a Machine Learning (ML) algorithm. It compares the performance of such a hybrid algorithm with an entirely classical algorithm. Quantum Machine Learning (QML) has been growing in recent years. Some studies suggest that QML may even provide a polynomial speed-up for data categorization compared to traditional ML. However, analyzing the benefits is not straightforward, as QML algorithms often rely on abstract, oracle (black-box) models that frequently rely on Quantum Random Access Memory (QRAM). Furthermore, loading classical data onto quantum registers limits the applicability of QML, imposing a bottleneck. We used the Swap Test to measure the overlap between two quantum states to achieve our objective. Then we replaced the classical distance metric in a distance-based machine learning algorithm with the quantum-based distance metric. Our research showed that the Swap Test could be used as a distance metric in classical algorithms, despite the fact that the results obtained are not better than the classical metrics. In the final discussion, we present some ways that can improve the obtained results. |
|---|---|
| Autores principais: | Sousa, Tomás Rodrigues Alves de |
| Assunto: | Quantum machine learning Machine learning Classification Clustering Swap test Distance metric Metrica de distancia |
| Ano: | 2023 |
| País: | Portugal |
| Tipo de documento: | dissertação de mestrado |
| Tipo de acesso: | acesso aberto |
| Instituição associada: | Universidade do Minho |
| Idioma: | inglês |
| Origem: | RepositóriUM - Universidade do Minho |
| Resumo: | This master’s thesis explores the advantages of using a quantum-based distance metric in a Machine Learning (ML) algorithm. It compares the performance of such a hybrid algorithm with an entirely classical algorithm. Quantum Machine Learning (QML) has been growing in recent years. Some studies suggest that QML may even provide a polynomial speed-up for data categorization compared to traditional ML. However, analyzing the benefits is not straightforward, as QML algorithms often rely on abstract, oracle (black-box) models that frequently rely on Quantum Random Access Memory (QRAM). Furthermore, loading classical data onto quantum registers limits the applicability of QML, imposing a bottleneck. We used the Swap Test to measure the overlap between two quantum states to achieve our objective. Then we replaced the classical distance metric in a distance-based machine learning algorithm with the quantum-based distance metric. Our research showed that the Swap Test could be used as a distance metric in classical algorithms, despite the fact that the results obtained are not better than the classical metrics. In the final discussion, we present some ways that can improve the obtained results. |
|---|