论文部分内容阅读
将相关性剪枝算法(CPA)和变学习率、附加动量方法结合提出了一种基于CPA的改进的BP神经网络剪枝算法.实验结果表明,改进的算法可以降低训练步数,加快神经网络的收敛速度,在测试数据集上的均方误差也得到了进一步的优化.
In this paper, an improved BP neural network pruning algorithm based on CPA is proposed by combining the pruning algorithm (CPA) with variable learning rate and additional momentum method.The experimental results show that the improved algorithm can reduce the training steps and speed up the neural network The convergence speed, the mean square error in the test data set has also been further optimized.