An incremental modular echo state network

被引：0

作者：

Li F.-J. ^{[1
,2
,3
]}

Qiao J.-F. ^{[1
,2
]}

机构：

[1] College of Electronic Information and Control Engineering, Beijing University of Technology, Beijing

[2] Beijing Key Laboratory of Computational Intelligence and Intelligent System, Beijing University of Technology, Beijing

[3] School of Mathematical Science, University of Ji'nan, Ji'nan

来源：

Kongzhi yu Juece/Control and Decision | 2016年 / 31卷 / 08期

关键词：

Echo state network; MSO problem; Prediction; Reservoir; Singular value;

D O I：

10.13195/j.kzyjc.2015.0913

中图分类号：

O1 [数学];

学科分类号：

0701 ; 070101 ;

摘要：

An incremental modular echo state network(IM-ESN) is proposed to solve the multiple superimposed oscillator(MSO) problem, which is difficult to be solved by conventional ESNs. The reservoir of IM-ESN is made up of sub-reservoirs which are mutually independent. The weight matrices of sub-reservoirs are designed via the singular value decomposition(SVD) method. Based on the block diagonal matrix theory, the generated sub-reservoirs are added to the existing network one by one. During the growth of the network, IM-ESN can guarantee the echo state property without posterior scaling of the weights. The experiment results on the MSO problem show that the IM-ESN can determine its network complexity to match the given applications automatically, with better prediction performance and robustness. © 2016, Editorial Office of Control and Decision. All right reserved.

引用

页码：1481 / 1486

页数：5

共 18 条

[1]

Jaeger H., Haas H., Harnessing nonlinearity: Predicting chaotic systems and saving energy in wireless communication, Science, 304, 5667, pp. 78-80, (2004)

[2]

Herbert J., The Echo State Approach to Analysing and Training Recurrent Neural Networks, 12, 8, pp. 1-43, (2001)

[3]

Li D., Han M., Wang J., Chaotic time series prediction based on a novel robust echo state network, IEEE Trans on Neural Networks and Learning Systems, 23, 5, pp. 787-799, (2012)

[4]

Qiao J.F., Bo Y.C., Han G., Application of ESN-based multi indices dual heuristic dynamic programming on wastewater treatment process, Acta Automatica Sinica, 39, 7, pp. 1146-1151, (2013)

[5]

Pan Y., Wang J., Model predictive control of unknown nonlinear dynamical systems based on recurrent neural networks, IEEE Trans on Industrial Electronics, 59, 8, pp. 3089-3101, (2012)

[6]

Skowronski M.D., Harris J.G., Automatic speech recognition using a predictive echo state network classifier, Neural Networks, 20, 3, pp. 414-423, (2007)

[7]

Chen Y., Hu J., Kasabov N., Et al., NeuCubeRehab: A pilot study for EEG classification in rehabilitation practice based on spiking neural networks, Springer Lecture Notes in Computer Science, 8228, 1, pp. 70-77, (2013)

[8]

Xue Y., Yang L., Haykin S., Decoupled echo state networks with lateral inhibition, Neural Networks, 20, 3, pp. 365-376, (2007)

[9]

Schmidhuber J., Wierstra D., Gagliolo M., Et al., Training recurrent networks by evolino, Neural Computation, 19, 3, pp. 757-779, (2007)

[10]

Zhi R., Gong Z.Q., Wang Q.G., Et al., The influence of global warming and extreme events on the correlation of the global temperature field, Acta Meteorologica Sinica, 70, 4, pp. 728-735, (2012)

← 1 2 →