Ali Wagdy Mohamed

To obtain the optimal set of features in feature selection problems is the most challenging and prominent problem in machine learning. Very few human-related metaheuristic algorithms were developed and solved this type of problem. It motivated us to check the performance of recently developed gaining–sharing knowledge-based optimization algorithm (GSK), which is based on the concept of gaining and sharing knowledge of humans throughout their lifespan. It depends on two stages: beginners–intermediate gaining and sharing stage and intermediate–experts gaining and sharing stage. In this study, two approaches are proposed to solve feature selection problems: FS-BGSK: a novel binary version of GSK algorithm that relies on these two stages with knowledge factor 1 and FS-pBGSK: a population reduction technique that is employed on BGSK algorithm to enhance the exploration and exploitation quality of FS-BGSK. The proposed approaches are checked on twenty two feature selection benchmark datasets from UCI repository that contains small, medium and large dimensions datasets. The obtained results are compared with seven state-of-the-art metaheuristic algorithms; binary differential evolution, binary particle swarm optimization algorithm, binary bat algorithm, binary grey wolf optimizer, binary ant lion optimizer, binary dragonfly algorithm and binary salp swarm algorithm. It concludes that FS-pBGSK and FS-BGSK outperform other algorithms in terms of accuracy, convergence and robustness in most of the datasets.

This paper introduces a novel differential evolution (DE) algorithm for solving constrained engineering optimization problems called (NDE). The key idea of the proposed NDE is the use of new triangular mutation rule. It is based on the convex combination vector of the triplet defined by the three randomly chosen vectors and the difference vectors between the best, better and the worst individuals among the three randomly selected vectors. The main purpose of the new approach to triangular mutation operator is the search for better balance between the global exploration ability and the local exploitation tendency as well as enhancing the convergence rate of the algorithm through the optimization process. In order to evaluate and analyze the performance of NDE, numerical experiments on three sets of test problems with different features, including a comparison with thirty state-of-the-art evolutionary algorithms, are executed where 24 well-known benchmark test functions presented in CEC’2006, five widely used constrained engineering design problems and five constrained mechanical design problems from the literature are utilized. The results show that the proposed algorithm is competitive with, and in some cases superior to, the compared ones in terms of the quality, efficiency and robustness of the obtained final solutions.

Proposing new mutation strategies to improve the optimization performance of differential evolution (DE) is an important research study. Therefore, the main contribution of this paper goes in three directions: The first direction is introducing a less greedy mutation strategy with enhanced exploration capability, named DE/current-to-ord_best/1 (ord stands for ordered) or ord_best for short. In the second direction, we introduce a more greedy mutation strategy with enhanced exploitation capability, named DE/current-to-ord_pbest/1 (ord_pbest for short). Both of the proposed mutation strategies are based on ordering three selected vectors from the current generation to perturb the target vector, where the directed differences are used to mimic the gradient decent behavior to direct the search toward better solutions. In ord_best, the three vectors are selected randomly to enhance the exploration capability of the algorithm. On the other hand, ord_pbest is designed to enhance the exploitation capability where two vectors are selected randomly and the third is selected from the global p best vectors. Based on the proposed mutation strategies, ord_best and ord_pbest, two DE variants are introduced as EDE and EBDE, respectively. The third direction of our work is a hybridization framework. The proposed mutations can be combined with DE family algorithms to enhance their search capabilities on difficult and complicated optimization problems. Thus, the proposed mutations are incorporated into SHADE and LSHADE to enhance their performance. Finally, in order to verify and analyze the performance of the proposed mutation strategies, numerical experiments were conducted using CEC2013 and CEC2017 benchmarks. The performance was also evaluated using CEC2010 designed for Large-Scale Global Optimization. Experimental results indicate that in terms of robustness, stability, and quality of the solution obtained, both mutation strategies are highly competitive, especially as the dimension increases.