CAJ | 학술논문

BP神经网络在人工神经网络中起着至关重要的作用，通过分析标准 BP 神经网络的基本算法，指出标准BP算法的一些不足，并针对这些不足提出了以一种以相对误差作为误差传递信号的新的改进方法。经试验证明：该方法大大提高了 BP神经网络预测结果的精度，同时这种新的改进思想也可以结合其他改进方法一起应用，以更大程度上地提高 BP神经网络的运算速度和预测精度。
BP신경망락재인공신경망락중기착지관중요적작용，통과분석표준 BP 신경망락적기본산법，지출표준BP산법적일사불족，병침대저사불족제출료이일충이상대오차작위오차전체신호적신적개진방법。경시험증명：해방법대대제고료 BP신경망락예측결과적정도，동시저충신적개진사상야가이결합기타개진방법일기응용，이경대정도상지제고 BP신경망락적운산속도화예측정도。
BP neural network plays a vital role in artificial neural networks.In this paper, through the analysis of the basic algorithm of standard BP neural network, and points out some shortcomings of the standard BP algorithm, to solve these problems we use a relative error as a new improved method of error transfer signal .The test proved that this method greatly improves the accuracy of BP neural network prediction, and this new and improved idea can also be applied to-gether with other improved methods to predict the computing speed and accuracy to a greater extent to improve BP neural network.