Soft Patterns Reduction for RBF Network Performance Improvement
Authors: 
Pawel Rozycki, Janusz Kolbusz, Oleksandr Lysenko, Bogdan M. WilamowskiAuthors Info & Claims
Pages 190 - 200
Abstract
Successful training of artificial neural networks depends primarily on used architecture and suitable algorithm that is able to train given network. During training process error for many patterns reach low level very fast while for other patterns remains on relative high level. In this case already trained patterns make impossible to adjust all trainable network parameters and overall training error is unable to achieve desired level. The paper proposes soft pattern reduction mechanism that allows to reduce impact of already trained patterns which helps in getting better results for all training patterns. Suggested approach has been confirmed by several experiments.
References
[1]
Wilamowski BM and Yu H Neural network learning without backpropagation IEEE Trans. Neural Netw. 2010 21 11 1793-1803
[2]
Wilamowski BM Neural network architectures and learning algorithms, how not to be frustrated with neural networks IEEE Ind. Electron. Mag. 2009 3 4 56-63
[3]
Hunter D, Yu H, Pukish MS, Kolbusz J, and Wilamowski BM Selection of proper neural network sizes and architectures-a comparative study IEEE Trans. Ind. Inf. 2012 8 228-240
[4]
Hohil ME and Liu D Solving the N-bit parity problem using neural networks Neural Netw. 1999 12 1321-1323
[5]
Wilamowski, B.M.: Challenges in applications of computational intelligence in industrial electronics. In: ISIE 2010, pp. 15–22, 4–7 July 2010
[6]
Fahlman, S.E., Lebiere, C.: The cascade-correlation learning architecture. In: Advances in Neural Information Processing Systems 2, pp. 524–532. Morgan Kaufmann, San Mateo (1990)
[7]
Lang, K.L., Witbrock, M.J.: Learning to tell two spirals apart. In: Proceedings of the 1988 Connectionists Models Summer School. Morgan Kaufman (1998)
[8]
Wilamowski, B.M., Korniak, J.: Learning architectures with enhanced capabilities and easier training. In: 19th IEEE International Conference on Intelligent Engineering Systems (INES 2015), 03–05 September, pp. 21–29 (2015)
[9]
Nguyen, G.H., Bouzerdoum, A., Phung, S.L.: Efficient supervised learning with reduced training exemplars. In: 2008 IEEE International Joint Conference on Neural Networks (IEEE World Congress on Computational Intelligence), Hong Kong, pp. 2981–2987 (2008)
[10]
Lozano, M.T.: Data reduction techniques in classification processes. Ph.D. Dissertation, Universitat Jaume I, Spain (2007)
[11]
Chouvatut, V., Jindaluang, W., Boonchieng, E.: Training set size reduction in large dataset problems. In: 2015 International Computer Science and Engineering Conference (ICSEC), Chiang Mai, pp. 1–5 (2015)
[12]
Kolbusz, J., Rozycki, P.: Outliers elimination for error correction algorithm improvement. In: CS&P Proceedings 24th International Workshop Concurrency, Specification & Programming, (CS&P 2015), vol. 2, pp. 120–129 (2015)
[13]
Rozycki P, Kolbusz J, Lysenko O, and Wilamowski BM Neural network training improvement by patterns removing Artif. Intell. Soft Comput. ICAISC 2017 2017 154-164
[14]
Yu H, Reiner P, Xie T, Bartczak T, and Wilamowski BM An incremental design of radial basis function networks IEEE Trans. Neural Netw. Learn. Syst. 2014 25 10 1793-1803
[15]
Xie, T.: Growing and learning algorithms of radial basis function networks. Ph.D. Dissertation, Auburn University, USA (2013)
Recommendations
Ml-rbf: RBF Neural Networks for Multi-Label Learning
Multi-label learning deals with the problem where each instance is associated with multiple labels simultaneously. The task of this learning paradigm is to predict the label set for each unseen instance, through analyzing training instances with known ...
GMDH-type neural network algorithm with a feedback loop for structural identification of RBF neural network
In this paper, a Group Method of Data Handling (GMDH)-type neural network algorithm with a feedback loop for structural identification of Radial Basis Function (RBF) neural network is proposed. In case of the GMDH-type neural network, the network ...
The Improvement of BP Artificial Neural Network Algorithm and Its Application
ICEE '10: Proceedings of the 2010 International Conference on E-Business and E-GovernmentThis paper is connected with the problem of selecting architectural parameters and learning rate of BP artificial neural network. The self-adapting algorithm of BP artificial neural network has been proposed, and the corresponding C language procedure ...
Comments
Please enable JavaScript to view thecomments powered by Disqus.Information & Contributors
Information
Published In
Jun 2018
795 pages
ISBN:978-3-319-91252-3
DOI:10.1007/978-3-319-91253-0
© Springer International Publishing AG, part of Springer Nature 2018.
Publisher
Springer-Verlag
Berlin, Heidelberg
Publication History
Published: 03 June 2018
Author Tags
Qualifiers
- Article
Contributors
Other Metrics
Bibliometrics & Citations
Bibliometrics
Article Metrics
- 0Total Citations
- 0Total Downloads
- Downloads (Last 12 months)0
- Downloads (Last 6 weeks)0
Reflects downloads up to 16 Dec 2024