基于神经网络的数字信号多分类识别方法

Multi-Classification Recognition Method for Digital Signals Based on Neural Network

文章针对数字信号的图像分类识别问题,提出了一种改进的神经网络算法,该算法利用随机梯度下降增量规则实现误差和上层输出共同影响权重的监督机制,采用softmax激活函数避免出现以很高的概率同时分到不同的类的问题,从而大大提高了识别准确率。

Aiming at the problem of image classification and recognition of digital signals, this paper proposed an improved neural network algorithm. The algorithm uses the random gradient descent incremental rule to realize the supervisory mechanism that errors and upper output affect the weight together. Softmax activation function is used to avoid the problem of classifying different signals at the same time with high probability. Thus, the recognition accuracy is greatly improved.