Performance of Quantum and Classical Algorithms in Supervised Machine Learning Training

Abstract

This article addresses the interdisciplinary theme of Quantum Computing with Machine Learning, two technologies potentially capable of making changes in how computing is performed, solving initially unsolvable problems. The focus of this research was Quantum Computing applications that result in computational performance gain in specific Machine Learning tasks. The objective is to analyze the feasibility of using quantum algorithms for Machine Learning. More specifically, to analyze which quantum algorithms can be applied to Machine Learning tasks, compared to classical algorithms, in the search for better performance. For the development of the research, a bibliographic review of quantum algorithms was carried out and, subsequently, the implementation and performance verification of the quantum algorithm QSVM and its corresponding classic version SVM, in supervised learning with the AD HOC and IRIS datasets.