Cubature Kalman Optimizer versus Teaching Learning Based Optimization: A Performance Comparison based on CEC2014 Test Suite

Abstract

This paper compares a new Cubature Kalman Optimizer performance against the Teaching Learning Based Optimization in solving the CEC2014 test suite. The Cubature Kalman Optimizer is inspired by the estimation algorithm named Cubature Kalman filter, while the Teaching Learning Based Optimization is inspired by the teaching-learning process in a classroom. Both algorithms can be characterized as a parameter-less nature. Graphical analysis based on convergence curve shows that Cubature Kalman Optimizer has better exploration than Teaching Learning Based Optimization in the first half of the total iteration that make it able to find better solution. On the other hand, for boxplot, both algorithms show comparative based on consistency. Meanwhile, statistical analysis shows that the Cubature Kalman Optimizer algorithm is a promising approach compared to Teaching Learning Based Optimization.