Nonlinear time series online prediction using reservoir Kalman filter

Authors:
Min Han;Yanan Wang
Affiliations:
School of Electronic and Information Engineering, Dalian University of Technology, Dalian, China;School of Electronic and Information Engineering, Dalian University of Technology, Dalian, China
Venue:
IJCNN'09 Proceedings of the 2009 international joint conference on Neural Networks
Year:
2009

Citing 9
Cited 1

Nonlinear dynamics, delay times, and embedding windows

Physica D
Real-time computing without stable states: a new framework for neural computation based on perturbations

Neural Computation
Online State--Space Modeling Using Recurrent Multilayer Perceptrons with Unscented Kalman Filter

Neural Processing Letters
Online stability of backpropagation-decorrelation recurrent learning

Neurocomputing
On the Kalman filtering method in neural network training and pruning

IEEE Transactions on Neural Networks
New results on recurrent network training: unifying the algorithms and accelerating convergence

IEEE Transactions on Neural Networks
A tighter bound for the echo state property

IEEE Transactions on Neural Networks
Support Vector Echo-State Machine for Chaotic Time-Series Prediction

IEEE Transactions on Neural Networks
Global Asymptotic Stability of Recurrent Neural Networks With Multiple Time-Varying Delays

IEEE Transactions on Neural Networks

Regularized variational bayesian learning of echo state networks with delay&sum readout

Neural Computation

Quantified Score

Hi-index	0.00

Visualization

Abstract

A novel online adaptive prediction method is proposed for complex time series. The KF is adopted in the high-dimension "reservoir" state space and directly updates the output weights of the echo state network (ESN) online. Compared with the expanded Kalman filter (EKF) algorithm of traditional recurrent neural networks, the reservoir KF method offers a implementation without the computation of numerical derivatives, so as to improve the prediction accuracy and extend the applications. Stability and convergence analysis of the proposed method is presented. Simulation examples demonstrate the validity of the proposed method.