Performance Analysis of Tree Seed Algorithm for Small Dimension Optimization Functions

Abstract

Tree-Seed Algorithm (TSA) simulates the growth of trees and seeds on a land. TSA is a method proposed to solve continuous optimization problems. Trees and seeds indicate possible solutions in the search space for optimization problems. Trees are planted in the ground at the beginning of the search and each tree produces several seeds during iterations. While the trees were selected randomly during seed formation, the tournament selection method was used and also hybridized by adding the C parameter, which is the acceleration coefficient calculated according to the size of the problem. In this study, continuous optimization problem has been solved by the hybrid method. First, the performance analyses of the five best known numerical benchmark functions have been done, in both TSA and hybrid method TSA with 2, 3, 4 and 5 dimensions, and 10-50 population numbers. After that, well-known algorithms in the literature like Particle Swarm Optimization (PSO), TSA, Artificial Bee Colony (ABC), Harmony Search (HS), as well as hybrid method TSA (HTSA) have been applied to twenty-four numerical benchmark functions and the performance analyses of algorithms have been done. Hopeful and comparable conclusions based on solution quality and robustness can be obtained with the hybrid method.