基于信息扩散原理运用人工神经网络识别股票级别

本文提出采用人工神经网络的方法对股票级别进行识别,以便辨识股票的优劣。首先建立股票识别的指标体系,对神经网络的训练样本中的各支股票的各项指标进行初步评级,然后用因素状态空间上信息扩散的方法对初步评级的结果进行优化处理,把处理后的结果作为神经网络训练样本的输入和输出因素,并利用BP算法进行网络训练,当计算达到要求的精度后即完成训练。经训练所得的神经网络即可用于识别股票级别。本文最后给出一个实例,具体说明运用信息扩散和人工神经网络的方法识别股票级别的过程,并对结果进行了分析讨论。采用本文所介绍的方法对股票级别进行识别,可以有效地减少主观因素及市场波动所带来的影响。

Using Artificial Neural Networks to Evaluate the Grading of Stock Based on the Principle of Information Spread

This paper proposed the method of applying artificial neural Networks to evaluate the grading of stock.At first,we establish the criteria system to evaluate the grading of stock,and then evaluate the stock index of training sample in ANN.Second,optimize the result based on information spread and use the optimized result to train neural networks with the back propagation.When arriving the necessary precision,the training will be stopped.The artificial neural networks may evaluate the grading of stock.In the end of this paper,we give a practical example and the detailed calculation steps,to explain this method and the grading results of stock.Using this method to evaluate the grading of stock will be helpful to reduce the influences of subjectivity and the market changes.