Research On Yellow Saddle Goatfish Algorithm Incorporating Improvement Strategies And Its Application

Intelligent optimization algorithm is an important heuristic algorithm based on fundamental theories or mathematical models in nature,with the advantages of insensitivity to initial conditions,simple operation and easy to understand and no need to limit the objective function,etc.It has become an effective technical tool for solving complex optimization problems,and highlight the efficiency and superiority in the application of optimization machine learning algorithms.However,intelligent optimization algorithms usually have disadvantages such as strong initialization randomness,poor local optimization,and solidification of key parameter settings,which easily lead to weak performance of algorithms escaping local extremes and deviations in the balance of global exploration and local search capabilities.Therefore,based on the research background of optimization theory and machine learning algorithm optimization,considering the local mining performance and global optimization capability of intelligent optimization algorithm,it is of great theoretical value and practical significance to study the improvement strategy of intelligent optimization algorithm and apply the improvement algorithm to the optimization problem of machine learning.In this paper,based on the traditional yellow saddle goatfish algorithm,in order to solve the problems of fixed step size parameter setting and easy to fall into local optimality,starting from the enhancement of search coverage and algorithm diversity,a dynamic step factor step-variable function is designed to realize the efficient and comprehensive optimization purpose of the algorithm in the search space.At the same time,the Fuch chaotic mapping theory is used to expand the local remining enhanced mode of the current optimal solution to complete the local optimization of the traditional algorithm.The mathematical experiment method is used to improve the algorithm's mathematics.Improved algorithm is tested and compared with the mathematical model using numerical experiments,and its improved optimization performance is systematically studied and analyzed.Finally,the improved yellow saddle goatfish algorithm is applied to the parameter combination optimization problem of the extreme learning machine algorithm in machine learning,and a comparison experiment is conducted with the extreme learning machine classification model of other population intelligence optimization algorithms to verify the efficiency and superiority of the yellow saddle goatfish algorithm incorporating the improved strategy to optimize the parameter model of the extreme learning machine.Research shown that the two improvement strategies increase the search coverage and optimization accuracy of the yellow saddle goatfish algorithm,and optimize the global exploration ability and local exploitation ability.The recursive mode of dynamic step size factor is helpful to improve the search efficiency and extend the search range of the algorithm,and the chaotic search mechanism completes the improvement of the local search performance of the yellow saddle goatfish algorithm with the superior chaotic property of Fuch mapping theory and better local convergence performance.The fusion method achieves a multi-round dynamic iterative balance between the global exploration and local search ability of the yellow saddle goatfish algorithm,which increases the diversity of the original algorithm to a certain extent,and the improved algorithm successfully circumvents the premature convergence phenomenon of the original algorithm with superior parallel iterative search performance and robustness.The optimized limit learning machine model based on the improved yellow saddle goatfish algorithm improves the classification accuracy of the limit learning machine while reducing the invalid iterations of the algorithm,so that the algorithm can effectively jump out of the local extremes.The optimized extreme learning machine model based on the improved yellow saddle goatfish algorithm has higher classification accuracy and better stability than other population intelligence algorithms,and has stronger optimal solution search ability,while still has better prediction performance under the worst extreme conditions.The research results provide an effective extension for the improvement of yellow saddle goatfish algorithm and provide theoretical support for the parameter optimization problem of extreme learning machines.The paper has 13 pictures,12 tables,and 82 references.