A Novel GA-LM Based Hybrid Algorithm

In order to improve the model´s learning capability and convergence rate, the GA and ANN are usually combined together. But the current combination ways have the insufficiencies of premature convergence, weak extensive ability etc. To overcome these shortcomings, we propose a new hybrid study algorithm-GALM, which uses the GA and LM in turn to optimize the neural network, we compare the GALM algorithm with other relevant algorithms through experimentation. The results indicate that our algorithm can effectively overcome the problem about falling into the local optimal solutions, and remarkably improved the network learning capability and the convergence rate.