A hybrid algorithm based on chips

A hybrid algorithm based on chips (HABC) is proposed to speed up the training of back-propagation neural networks, and to improve the performances of neural networks. The algorithm divide the training of neural networks into many training chips, and an improved BP algorithm based on magnified error signal is performing on those chips. The genetic algorithm is introduced to optimize the results of chips training when a chip training is accomplished. Then the next chip training is carrying out on the optimized result. Therefore, the HABC obtains the ability of searching the global optimum solution relying on these optimal operations, and it is easy to be parallel processed. The simulation experiments show that this algorithm can effectively avoid failure training caused by randomizing the initial weights and thresholds, and solve the slow convergence problem resulted from the Flat-Spots when the error signal becomes too small.