Stefano Biondi M.Sc. Thesis in Data Science Repository
On one hand, recent computational developments allowed Machine Learning and Deep Learning techniques to achieve a huge success. On the other hand, quantum computation can outperform classical computers for specific classes of problems. In this scenario, current research is exploring hybrid quantum-classical systems able to leverage the strength of both systems. The objective of this thesis is to review the two fields and describe some applications of hybrid systems with variational quantum circuits (VQC) applied to classical machine learning and deep learning algorithms. We show the strengths and weaknesses of this new approach. We also propose a new way to reduce parameters in the construction of VQC architectures and demonstrate that, for a specific problem that is state-of-the-art in the field, it reduces the number of parameters by 8 times without losing performance. Finally we give a proof-of-principle implementation of the proposed architecture running on quantum hardware from Rigetti.