基于状态转移算法的极限学习机

In order to solve the problems of extreme learning machine (ELM), occupying more computing resources and low accuracy during model training, an extreme learning machine based on the state transition algorithm (STA) was proposed to improve the calculation efficiency of the algorithm and the accuracy of the model. Taking advantage of the global search feature of the state transition algorithm, the algorithm was arranged to solve the linear equations, obtain the output weight matrix of the extreme learning machine and complete the modeling. Compared with extreme learning machine and other mainstream algorithms on classification and regression data sets, the proposed algorithm can realize high model accuracy with fewer hidden layer nodes and achieve better learning accuracy. The high-performance modeling method can make up for the deficiencies of the extreme learning machine.