Research on support vector machine optimization based on improved quantum genetic algorithm

Support vector machine (SVM) is one of the classical machine learning algorithms. It is widely used and researched for its generalizability and low sample data requirements. However, classical SVM often suffers from problems such as slow solving speed and insufficient accuracy. Quantum genetic algorithm (QGA), which is based on quantum computing principle, is characteristic of faster solving speed and better accuracy than that of classical genetic algorithm, but it is also deficient in easiness of resorting to local optimal solution and insufficient convergence rate in the face of complex problems. In this paper, we propose an improved quantum genetic algorithm (IQGA), which designs a crossover evolution strategy and dynamic rotation angle to avoid local optima. It also adds a quantum convergence gate to address the issue of local convergence. The improved quantum genetic algorithm is applied to SVM parameter optimization, thus its superiority through experimentation and analysis is demonstrated. The results indicate that the model based on improved quantum genetic algorithm support vector machine (IQGA-SVM) has higher prediction accuracy and faster convergence rate compared to back-propagation neural network, classical genetic algorithm support vector machine (GA-SVM) and quantum genetic algorithm support vector machine (QGA-SVM) models.