An exchange rate forecasting method based on probabilistic neural network

Probabilistic neural network architecture is the most important features of real-time network training, the probabilistic neural network system suitable for use in real-time. Probabilistic neural networks and decision-making side of the scope of the problem size can be based on different needs to be adjusted, making surface can approximate Bayesian decision-making. Probabilistic neural network and noise tolerance for error is high. In the sparse sample space, the smoothing parameter based on the needs of the problem, adjust the parameters of any size, without the need for re-training.