Improvement Of Artificial Bee Colony Algorithm And Its Application In Protein Classification

Swarm intelligent optimization algorithm is inspired by the living habits of gregarious biology,and currently it has become a hotspot to solve complex optimization problems.Artificial bee colony(ABC)algorithm,which is inspired by the foraging behavior of honey bee swarm,is a novel swarm intelligent optimization algorithm.Compared with other swarm intelligence optimization algorithms,ABC algorithm has the advantages of less control parameters,strong ability of global exploration,and easy to implement.It has been widely used in many fields such as economy,society,science and medicine.However,the algorithm also has some problems,such as poor ability of local exploitation,premature convergence and low accuracy of the objective function value,so the research on the basic artificial bee colony algorithm still has a broad prospect.Protein is considered to be the basic element in life,and it has various functions to sustain life,that also makes proteomics a very important research field in modern bioinformatics.Proteins can be divided into different categories according to their functions,and proteins in the same category have similar structures and properties.it is of great significance to study the classification of proteins to determine their functions.Based on the performance analysis and theoretical research of artificial bee colony algorithm,this paper proposes an artificial bee colony algorithm combined with difference operators,and discusses the application of artificial bee colony algorithm in protein classification.The work content of this article mainly includes the following aspects:(1)The Basic Theory of Artificial Bee Colony Algorithm.This article expounds the research background and research status of artificial bee colony algorithm,deeply studies the biological background,basic principles and algorithm framework of artificial bee colony algorithm,analyzes the advantages and disadvantages and application scope of artificial bee colony algorithm and other intelligent optimization algorithms.The time complexity of the algorithm is discussed,and four different benchmark functions are selected to test and analyze the global convergence of the algorithm.(2)Artificial Bee Colony Algorithm Combining Difference Operators.Artificial bee colony(ABC)algorithm does well in exploration but badly in exploitation.Unlike ABC,DE tends to exploit well but weakly in exploration.Therefore,the combination algorithm of ABC and DE is proposed,named AMDABC(adaptive modified differential operators based artificial bee colony).AMDABC follows the framework of artificial bee colony algorithm,including the phase of employing bees,onlooker bees and scout bees.This paper introduces two DE operators,operators of CoDE and operators of JADE in the employing bee phase and gives two control parameters.According to the value of the control parameter,the CoDE operator,JADE operator or ABC search equation are performed adaptively and alternately to achieve the balance between global exploration ability and local exploitation ability.In the onlooker bee phase,the JADE difference operator is also used to generate candidate solutions to be a better way to solve the problem of weak exploitation ability of ABC algorithm.Experiments on 19 benchmark functions show that the performance of AMDABC is superior to that of ABC,DE and hybrid algorithm of ABC and DE.(3)Artificial Bee Colony Algorithm optimization SVM study on MHC class I protein classification.The penalty factor and kernel parameters of SVM affect the classification performance of SVM,so the parameters can be optimized by optimization algorithms to improve the classification performance of SVM.This paper proposes an artificial bee colony algorithm to optimize the classification algorithm of SVM parameters(ADA-SVM).In this method,the artificial bee colony algorithm is used to optimize the penalty factors and kernel parameters.The position of the food source of the artificial bee colony algorithm represents the penalty factor and kernel parameters of the SVM,and the fitness function is expressed by the classification accuracy.The process of searching for the optimal position of the food source is the process of finding the optimal parameters by the SVM.The optimal preset parameter ranges of penalty factors and kernel parameters are determined through experiments on 4 classic data sets in UCI database,and compared with other parameter-optimized SVM algorithms.The experimental results show the classification of the ADA-SVM algorithm proposed in this paper Better performance.Finally,the ADA-SVM algorithm proposed in this paper is used for MHC class I protein classification,and the RBF radial basis kernel function is determined as the kernel function by experimental methods.Experimental results with other classification algorithms show that ADA-SVM has better classification performance,the classification accuracy of MHC class I protein can reach 98.45%,the classification effect is better than other algorithms,and proves the effectiveness of ADA-SVM classification method.