Quantum enhancements for machine learning based on a probabilistic quantum memory

Abstract

Quantum machine learning arises from the interaction of fields of machine learning and quantum computing. Machine learning is a branch of artificial intelligence relevant in many areas. It provides computers the ability to learn autonomously from experience. Quantum computing, on the other hand, is a different computational paradigm. The processing of information and communication in a quantum computer makes use of the principles and properties of quantum mechanics. With this, it is possible to achieve computational effects that cannot be efficiently reached classically. Quantum computing raises new possibilities through promising approaches that make use of these effects. In fact, proposed quantum algorithms demonstrate their potential in outperforming classical algorithms in some tasks. The present work aims to contribute with the field of quantum machine learning. In order to do so, the use and applications of a quantum probabilistic memory as a tool to propose improved machine learning algorithms is investigated. Here, the quantum memory is used to develop improved procedures for tasks such as cross-validation, and the selection and evaluation of artificial neural network architectures. In addition, a weightless neural network model using the probabilistic quantum memory was evaluated and improved.