Quantum Fuzzy Support Vector Machine for Binary Classification

In the objective world, how to deal with complexity and uncertainty of big data efficiently accurately has become premise key machine learning. Fuzzy support vector (FSVM) not only deals classification problems for training samples fuzzy information, but also assigns a membership degree each sample, allowing different contribute differently in predicting an optimal hyperplane separate two classes maximum margin, reducing effect outliers noise, Quantum computing super parallel capabilities holds promise faster algorithmic processing data. However, FSVM quantum are incapable dealing efficient accurate manner. This paper research propose (QFSVM) algorithm based on fact that can process large amounts is easy problems. The central idea proposed use solving linear systems equations (HHL algorithm) least-squares method solve quadratic programming problem FSVM. determine whether sample belongs positive or negative class while achieving good generalization performance. Furthermore, this applies QFSVM handwritten character recognition demonstrates be run computers, achieve characters. When compared FSVM, QFSVM’s computational decreases exponentially number samples.