Financial data simulation using M-PHONN model
Zhang Ming, Bo Lu · 2002
A new model, called Multi-Polynomial Higher Order Neural Network (M-PHONN), has been developed. Using Sun workstation, C++, and Motif, a M-PHONN simulator has been built as well. Real world data always can not be simply simulated very well by single polynomial function. So the ordinary higher order neural networks could fail to simulate such complicated real world data. But M-PHONN model can simulate multipolynomial functions, it makes M-PHONN model can achieve more accuracy for real world data simulation. The comparison experiments between M-PHONN and ordinary higher order neural network also shows that M-PHONN always can have 2-50% more accuracy than ordinary higher order neural networks.