AN IMPROVED CUCKOO SEARCH ALGORITHM FOR LINEAR LEAST SQUARES
MOHAMED ABDEL-BASET *
Department of Operations Research, Faculty of Computers and Informatics, Zagazig University, El-Zera Square, Zagazig, Sharqiyah, Egypt.
IBRAHIM M. HEZAM
Department of Computer, Faculty of Education, Ibb University, Ibb City, Yemen.
*Author to whom correspondence should be addressed.
Abstract
In this paper, we propose a new algorithm that encompasses the features of cuckoo search and Newton method. It combines the fast convergence and the global search of the Cuckoo Search (CS), and the strong local search ability of Newton method (NM). The Compound Algorithm (CSNM) will be used to solve linear least squares problem. The linear least squares problem is transformed into an optimization problem, where the fitness function is defined as the sum of squared residuals. The proposed algorithm retained the global search capability with more accurate and faster convergence. The experimental results show that the proposed algorithm proved to be superior in convergence efficiency and computational accuracy.
Keywords: Cuckoo search algorithm, meta-heuristics, optimization, newton method, linear least squares