Quantum Machine Learning Applied To Natural Language Processing

Voir Plus Voir Moins

Afficher le Résumé

A social networking platforms continue to grow, users increasingly share their thoughts and opinions through various forms of text. Despite this active engagement, sentiment analysis remains a significant challenge due to the vast volume of textual data generated from diverse sources. This task has gained considerable attention within the field of Natural Language Processing (NLP), as it provides valuable insights into public opinion, user experiences, and customer feedback. Meanwhile, quantum mechanics has reshaped our understanding of the world, and one of its emerging branches Quantum Machine Learning (QML) has shown promising theoretical results. However, the application of QML in sentiment analysis is still limited and remains largely at an experimental or theoretical stage. In this thesis, we evaluate one such QML approach: Quantum Convolutional Neural Networks (QCNNs) for movie review classification. QCNNs are quantum neural network architectures inspired by classical Convolutional Neural Networks (CNNs), but implemented entirely using parameterized quantum circuits. We investigate how different QCNN configurations with different number of qubits and structure of their parameterized circuits affect sentiment classification performance on both artificial and real movie review datasets.

Télécharger Document