Abstract
The influence of training data set on the supervised learning performance of artificial neural network (ANN) is studied in detail in this paper. First, some illustrative experiments are conducted, which verify that different training data set can lead to different supervised learning performance of ANN; secondly, the necessity of carrying data preprocessing to training data set is analyzed, and how training data set affect the supervised learning is summarized; at last, the existing methods about improving performance of ANN by using high-quality training data are discussed.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Preview
Unable to display preview. Download preview PDF.
References
Philip, N.S.: Optimal selection of training data for the difference boosting neural networks. In: Proc. of iAstro, Nice, France, October 16-17, pp. 1–9 (2003)
Liu, G., Zhang, H.G., Guo, J.: The influence of different training samples to recognition system. Chinese Journal of Computers 28(11), 1923–1928 (2005)
Zhu, X.J.: Semi-supervised learning literature survey, Technical report 1530, Computer Science. University of Wisconsin Madiso (2008)
Pereira, C.E.: Pereira Learning Vector Quantization with Training Data Selection. IEEE Trans. Pattern and analysis and machine intelligence 28(1), 157–162 (2006)
Vijayakumar, S., Sugiyama, M., Ogawa, H.: Training Data Selection for Optimal Generalization with Noise Variance Reduction in Neural Networks. In: Processings of Neural Nets WIRN Vietri-1998, Salerno, Italy, pp. 153–166 (1998)
Fukumizu, K.: Stastical active learning in multilayer perceptrons. IEEE Transactions on Neural Networks 11(1), 17–26 (2000)
Wang, S.B., Chai, Y.L., Liang, X.P.: Comparison of methods to produce sample points in training ANN. Journal of Computers Zhengzhou University (Engineering Science) 24(1), 63–69 (2003)
He, Y., Xiang, L.G.: Study and application of BP neural network model based on fuzzy clustering. Systems Engineering - Theory and Practice (2), 79–82 (2004)
Hao, H.W., Jiang, R.R.: Training sample selection method for neural networks based on nearest neighbor rule. Acta Automatica Sinica 33(12), 1247–1251 (2007)
Ma, X., Chen, X.C., Wang, S.B.: Application of the uniform design to the optimal selection of samples for RBF neural networks. PR & AI 18(2), 252–255 (2005)
Zhang, L., Guo, J.: A method for the selection of training samples based on boundary samples. Journal of Beijing Univerisity of Posts and Telecommunications 29(4), 77–80 (2006)
Hara, K., Nakayama, K.: A training method with small computation for classification. In: Proceedings of the IEEE-INNS-ENNS International Joint Conference on Neural Networks (IJCNN 2000), pp. 3543–3547 (2000)
Schohn, G., Cohn, D.: Less is more: active learning with support vector machines. In: Proceedings of 17th International Conference on Machine Learning, pp. 839–846 (2000)
Guan, D., Yuan, W., Lee, Y.-K., Gavrilov, A., Lee, S.: Improving supervised learning performance by using fuzzy clustering method to select training data. Journal of Intelligent & Fuzzy Systems 19(4), 321–334 (2008)
Li, C.H., Liu, C.M., Cai, G.: Approach to eliminating morbid samples in forward neural networks. Journal of Jilin University (Information Science Edition) 27(5), 514–519 (2009)
Ai, J.J., Zhou, C.G., Gong, C.C.: Algorithm of voting to eliminate morbid samples in forward feed neural networks. Mini - Micro System 11(11), 1371–1374 (2002)
Muhlenbach, F., Lallich, S., Zighed, D.A.: Identifying and handling mislabeled instances. Journal of Intelligent Information Systems 22(1), 89–109 (2004)
Author information
Authors and Affiliations
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2011 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Zhou, Y., Wu, Y. (2011). Analyses on Influence of Training Data Set to Neural Network Supervised Learning Performance. In: Jin, D., Lin, S. (eds) Advances in Computer Science, Intelligent System and Environment. Advances in Intelligent and Soft Computing, vol 106. Springer, Berlin, Heidelberg. https://doi.org/10.1007/978-3-642-23753-9_4
Download citation
DOI: https://doi.org/10.1007/978-3-642-23753-9_4
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-642-23752-2
Online ISBN: 978-3-642-23753-9
eBook Packages: EngineeringEngineering (R0)