A Survey of Optimization Algorithms in Support Vector Machine

Optimization algorithm solving Lagrangian multipliers is the key of training SVM, determining the performance of SVM, affecting practical applications of SVM in various fields widely. Some kinds of optimization algorithms in SVM of overseas are introduced. We classify the optimization algorithms into two kinds : 1. the algorithms based on Osuna's decomposition strategy; 2. The iterative algorithms based on the changes of SVM formulation proposed by O. L. Mangasanan. We also analyze the characteristics of various optimization algorithms in SVM,and predicting the trend of research on optimization algorithm in SVM.