Robust Granular Neural Networks, Fuzzy Granules and Classification
We introduce a robust granular neural network (RGNN) model based on the multilayer perceptron using back-propagation algorithm for fuzzy classification of patterns. We provide a development strategy of the network mainly based upon the input vector, linguistic connection weights and target vector. While the input vector is described in terms of fuzzy granules, the target vector is defined in terms of class membership values and zeros. The connection weights among nodes of RGNN are in terms of linguistic variables, whose values are updated by adding two linguistic hedges. The updated linguistic variables are called generalized linguistic variables. The node functions of RGNN are defined in terms of linguistic arithmetic operations. We present the experimental results on several real life data sets. Our results show that the classification performance of RGNN is superior to other similar type of networks.
Keywordsgranular computing linguistic variable linguistic arithmetic granular neural networks fuzzy classification
Unable to display preview. Download preview PDF.
- 1.Yao, Y.Y.: Granular Computing: basic issues and possible solutions. In: 5th Joint Conference on Information Sciences, pp. 186–189 (2001)Google Scholar
- 4.Dick, S., Tappenden, A., Badke, C., Olarewaju, O.: A Novel Granular Neural Network Architecture. In: Annual Meeting of the North American Fuzzy Information Processing Society - NAFIS, North America, pp. 42–47 (2007)Google Scholar
- 6.Liu, C., Shindhelm, A., Li, D., Jin, K.: A numerical approach to linguistic variables and linguistic space. In: IEEE Int. C. Fuzzy Systems, pp. 954–959 (1996)Google Scholar
- 7.Klir, G.J., Folger, T.: Fuzzy sets, Uncertainty and Information. Addison Wesley, Reading (1989)Google Scholar
- 9.Pal, S.K., Mitra, S.: Multilayer Perceptron, Fuzzy Sets, and Classification. IEEE Transactions on Neural Networks 3(5) (1992)Google Scholar