Training Multi-Layer Perceptrons with the Unscented Kalman Particle Filter

Yan Hui Xi; Hui Peng

doi:10.4028/www.scientific.net/AMR.542-543.745

Paper Titles

Study on Nozzle Array Structure Fluidic Gyroscope
p.727

Study on Improving Resolution of Fluidic Gyroscope
p.731

Uncertainty Analysis and Simulation of Coordinate Measuring Definition
p.735

Review of Acupoints Positioning and Tracking Technology for Chinese Massage Robot
p.741

Training Multi-Layer Perceptrons with the Unscented Kalman Particle Filter
p.745

Research on the Vessels' Routeing System and Ship Reporting System off Cheng Shanjiao
p.749

Study on the Speed Control System of Hydraulic Pump-Motor Based on PID Neural Network
p.759

A Study of the Effect of Absorbing Materials in the UHF Sensor
p.765

An Offset Reduction Technique for Latch Type Sense Amplifier in High Performance and High Density SRAM
p.769

HomeAdvanced Materials ResearchAdvanced Materials Research Vols. 542-543Training Multi-Layer Perceptrons with the...

Training Multi-Layer Perceptrons with the Unscented Kalman Particle Filter

Abstract:

Many Bayesian learning approaches to multi-layer perceptrons (MLPs) parameters optimization have been proposed such as the extended Kalman filter (EKF). In this paper, a sequential approach is applied to train the MLPs. Based on the particle filter, the approach named unscented Kalman particle filter (UPF) uses the unscented Kalman filter as proposal distribution to generate the importance sampling density. The UPF are devised to deal with the high dimensional parameter space that is inherent to neural network models. Simulation results show that the new algorithm performs better than traditional optimization methods such as the extended Kalman filter.

You might also be interested in these eBooks

View Preview

Info:

Periodical:

Advanced Materials Research (Volumes 542-543)

Pages:

745-748

DOI:

https://doi.org/10.4028/www.scientific.net/AMR.542-543.745

Citation:

Cite this paper

Online since:

June 2012

Authors:

Yan Hui Xi, Hui Peng

Keywords:

Bayesian Methods, Multi-Layer Perceptrons, Unscented Kalman Particle Filter

Export:

RIS, BibTeX

Price:

Permissions CCC:

Request Permissions

Permissions PLS:

Request Permissions

Сopyright:

Citation:

References

[1] S. Singhal, and L. Wu: Advance in Neural Information Processing Systems 1 (Morgan Kaufmann Publishers Inc., USA 1989).

Google Scholar

[2] Y. Iiguni, H. Sakai and H. Tokumaru: IEEE Transactions on signal processing, vol. 40 (1992) No.4, pp.959-966.

Google Scholar

[3] G. V. Puskorius and L. A. Feldkamp: Proceedings of International Joint Conference on Neural Networks ( Seattle, Washington, US, 1991), pp.307-312.

Google Scholar

[4] Information on http://www-sigproc.eng.cam.ac.uk/smc/papers.html.

Google Scholar

[5] J. F. G. de Freitas, M. Niranjan and A. Gee: IEEE International Conference on Acoustics and Signal Processing (Arizona, USA, 1999), vol. 2, pp.1057-1060.

Google Scholar

[6] J. F. G. de Freitas, et al: Neural Computation, vol. 12 (2000) No.4, pp.955-993.

Google Scholar

[7] R. H. Zhan and J. W. Wan: IEEE Signal Processing Letters, vol. 13 (2006) No.7, pp.445-448.

Google Scholar

[8] M.V. Rajesh, et al: Control and Decision Conference (Guilin, China, 2009), p.1477 – 1482.

Google Scholar

[9] X. L. Deng and P. F. Zhou: International Conference on Computational Intelligence and Natural Computing (Wuhan, China, 2009), pp.30-33.

Google Scholar

[10] G. Kitagawa: Journal of Computational and Graphical Statistics, vol. 5 (1996) No.1, pp.1-25.

Google Scholar

[11] R. van der Merwe, et al: Advances in Neural Information Processing Systems 13 (Massachusetts Instituite of Technology Press, USA 2001).

Google Scholar