Reconstructions and predictions of nonlinear dynamical systems by Rao-Blackwellised sequential Monte Carlo
Takao Soma, Kuniaki Yosui, Takashi Matsumoto · 2003 IEEE International Conference on Acoustics, Speech, and Signal Processing, 2003. Proceedings. (ICASSP '03). · 2003
Sequential Monte Carlo (SMC) is a powerful sampling based inference/learning algorithm for Bayesian scheme. The purpose of this paper is two fold. It first attempts to reconstruct and predict nonlinear dynamical systems from one dimensional data which arrives in a sequential manner instead of batch manner. Second purpose is to test the performance of the Rao-Blackwellisation in reconstructing and predicting nonlinear dynamical systems. We demonstrate that Rao-Blackwellised sequential Monte Carlo (RBSMC) on a chaotic time series prediction problem outperforms generic SMC.