An Optimized Extreme Learning Machine Based On Improved Dung Beetle Algorithm

Abstract

In order to improve the performance of the extreme learning machine, the improved dung beetle algorithm was used to optimize the parameters of the extreme learning machine. Firstly, the PWM chaotic map was called on the DBO algorithm to change its initialization process, enhance the diversity of the population, and accelerate the convergence speed. The Levy strategy is introduced into the updated formula at the location of the thief dung beetle to enhance the ability of the algorithm to jump out of the local optimum. Then, a spiral search strategy is introduced, which can be used to change the position, which ensures the convergence speed of the algorithm and can also increase the diversity of the population. Finally, by comparing the algorithm performance of the sparrow algorithm, the dung beetle algorithm, the gray wolf algorithm and other algorithms on 10 benchmark functions, the results show that the proposed algorithm has a faster convergence speed and is not easy to fall into the local optimum. Finally, the improved algorithm is used to optimize the input weight threshold of the extreme learning machine, and the modeling accuracy of the optimized extreme learning machine is improved through the simulation test of the benchmark dataset.