Time series prediction using graph model
Qinkun Xiao, Yang Si · 2017
A time series prediction model is proposed based on combination of the long-short time memory neural network model and the Dynamic Bayesian Network (DBN). Our contribution includes (1) An optimal estimation theorem is proposed and proved. We can get the optimal prediction estimation based on the proposed estimation theorem. (2) Based on the theorem, the recursion-based graph model is used to enhance prediction performance through probability inference. A new graph model called LSTM-DBN generated from a combination of LSTM prediction and DBN is developed to predict series data. The simulation results show that the model is premising.