Mechanism Improvement And Application Research Of Lions Algorithms

For a long time,swarm intelligent optimization algorithm,as a star of evolutionary computing technology,plays a role in both theoretical analysis and practical application.Compared with gradient-based method,swarm intelligent optimization algorithms do not need continuous and differentiable operation environment,and can adapt to the most of optimization problems.Especially for some NP hard problems,it can also obtain better optimization results.By imitating the change of animal population in nature,swarm intelligent optimization algorithm which is an important kind of swarm intelligent optimization algorithm,provides many new solutions for optimization problems.In recent years,by simulating the evolution or predation of lions,several lions algorithms based on different optimization mechanisms have been proposed.Among them,there are two kinds of lions algorithms,which are Lion Pride Optimization Algorithm based on genetic evolution process and Lion Swarm Optimization Algorithm based on hunting cooperation mechanism.These two algorithms have attracted widely attention of scientific research because of their superior optimization ability and perfect information exchange mechanisms.However,these two algorithms have their own shortcomings.In this paper,considering the problems of high time complexity and being easy to fall into local optimization,we improve their optimization mechanisms,and apply the improved algorithms to grayscale image segmentation and breast diseases detection.The experimental results show the validity,robustness and adaptability of these improved lions algorithms.Firstly,we introduce the theoretical mechanisms and mathematical models of two kinds of lions algorithms in detail.The Lion Pride Optimization Algorithm is similar to the Genetic Algorithm,with good local search ability and efficient mutation factor.Therefore,Lion Pride Optimization Algorithm has good results but with relatively high time and space complexity.Lion Swarm Optimization Algorithm based on hunting is similar to the Artificial Bee Colony Algorithm,which has high utilization rate of effective information among populations and can quickly converge,but it is easy to fall into local extremaIn view of the relatively high time complexity,we integrate the differential mechanism into the process of updating and variation of Lion Pride Optimization Algorithm,which improves the population diversity and convergence speed.Then,we propose a grayscale image segmentation method based on Fuzzy C-Means and differential-based Lion Pride Optimization Algorithm.In the grayscale image segmentation based on fuzzy clustering,the setting of the initial clustering centers is random,which has a great influence on the result of segmentation.Therefore,in this paper,the initial clustering centers are selected by the differential-based Lion Pride Optimization Algorithm,which can improve the effect of grayscale image segmentation.Then,given that Lion Swarm Optimization Algorithm is easy to fall into local extrema,the mechanism in the updating process of cubs is improved.What's more,a chemotaxis mechanism is added to improve the local optimization ability of Lion Swarm Optimization Algorithm.And a chemotaxis-based Lion Swarm Optimization Algorithm is proposed,which has improved optimization accuracy.Based on the characteristics of fast convergence and high precision of Lion Swarm Optimization Algorithm,we apply it to hyper-parameters selection of RBF kernel function in Support Vector Machine.In the process of breast diseases detection,Support Vector Machine with hyper-parameters selection based on chemotaxis-based Lion Swarm Optimization Algorithm has achieved good results.Finally,we summarize and analyze the mechanisms improvement and application research of lions algorithms.What's more,we also make prospects for the future of the proposed algorithms and application scenarios.