Hybridization of Adaptive Differential Evolution and Two Local Search Techniques with a Restart Strategy: Some Preliminary Experimental Results

DOI: 10.5176/2251-2179_ATAI12.24

Authors: Rashida Adeeb Khanum, Muhammad Asif Jan

Abstract: Classical optimization methods like Steepest Descent Method (SDM) and Broyden-Fletcher-Goldfarb-Shanno (BFGS) can solve many optimization problems. However, in case of new complex optimization problems (e.g., CEC2005 and CEC2010) with many local optima, they usually do not find the global optimum, because they get trapped in local optimum. This creates a gap for hybridizing them with other search techniques like Evolutionary Algorithms (EAs). Differential Evolution (DE) is the most recent EA which is a global optimizer and is very good in exploration, but is not as good as local search (LS) methods in exploitation. This paper introduces a new hybrid algorithm which combines Adaptive Differential Evolution (JADE) algorithm and two gradient-based classical methods, SDM, an inexpensive LS and BFGS, an expensive LS, for continuous unconstrained global optimization problems. The proposed algorithm is known as hybridJADE. hybridJADE is tested for 1000 dimensional benchmark functions from the CEC2010 collection. The obtained results are compared with those obtained by state-of-the-art SDENS and DASA algorithms.