Description: It is an implementation of hierarchical (a.k.a. multi-scale) Kalman filter using belief propagation. The model parameters are estimated by expectation maximization (EM) algorithm. In this implementation, we considered two time series with different frequencies. The messages between high and low frequency signals are combined to improve the estimation and prediction.
To Search:
File list (Check if you may need any files):
license.txt
multiScale_KalmanFilter
.......................\abstractsubformat-BME2012.pdf
.......................\Demo_KalmanFilter_BP_multiScale.m
.......................\EMConvergence.m
.......................\EMForMultiScaleKalmanFilter.m
.......................\eyeInf.m
.......................\GaussianDivision.m
.......................\GaussianMultiply.m
.......................\gaussian_prob.m
.......................\getCurrentZID.m
.......................\initializeMessage.m
.......................\inv_s.m
.......................\kalmanFilter_MultiScale.m
.......................\kalmanSmoother_MultiScale.m
.......................\linspaceInt.m
.......................\timeSeriesDataGeneration.m
.......................\updateMSGInLevel2.m
.......................\updateMSGToLevel2Backward.m
Download