Training Fuzzy Neural Networks with Social Spider Algorithm

Abstract: Inspired from the biological nervous systems artificial neural networks (ANNs) are simplified mathematical models of human brain. They automatically perform abilities of human such as learning, making new inferences and they are very successful in solving complex nonlinear problems. ANNs learn from examples and learning is achieved by training the network. Many algorithms and methods are developed so far for this purpose. Among these methods and algorithms conventional algorithms are based on mathematical computation, slow and often trapped into local optima. Fuzzy neural networks and meta-heuristic algorithms are widely used in literature to overcome these disadvantages with their remarkable properties like learning accuracy, speed and generalization capability. Fuzzy neural systems incorporate the fuzzy logic into the ANNs to deal with cognitive uncertainties like human brain. This can be done by fuzzifying input layer and inserting fuzzy logic operators into neural networks. In this study artificial neural networks and fuzzy neural networks are trained with social spider algorithm which is a novel meta-heuristic algorithm inspired from the social spider behaviors. Performances of the two algorithms are tested and compared on three classification benchmark datasets in machine learning area. Fuzzy neural networks outperform the artificial neural networks with higher classification accuracies.

Keywords: Fuzzy neural networks, Artificial neural networks, Social spider algorithm, Classification

Publication details: Conference abstract presented at the The 5th International Fuzzy Systems Symposium (FUZZYSS'17), 14-15 October 2017, Ankara, Türkiye. Published in: The 5th International Fuzzy Systems Symposium (FUZZYSS'17): Programme and Abstracts, pp. 87. Publisher: Fuzzy Systems Association & TOBB University of Economics and Technology.

How to cite / Kaynak gösterimi

APA 7: Gülmez, B., Kulluk, S., & Torun, H. (2017, October). Training Fuzzy Neural Networks with Social Spider Algorithm [Conference abstract]. In Ç. H. Aladağ, D. Alptekin, & B. Alptekin (Eds.), The 5th International Fuzzy Systems Symposium (FUZZYSS'17): Programme and Abstracts (p. 87). Fuzzy Systems Association & TOBB University of Economics and Technology. The 5th International Fuzzy Systems Symposium (FUZZYSS'17), Ankara, Türkiye. https://doi.org/10.5281/zenodo.23036889

IEEE: B. Gülmez, S. Kulluk, and H. Torun, "Training Fuzzy Neural Networks with Social Spider Algorithm," in The 5th International Fuzzy Systems Symposium (FUZZYSS'17): Programme and Abstracts, Ankara, Türkiye, Oct. 2017, p. 87, doi: 10.5281/zenodo.23036889.

Chicago: Gülmez, Burak, Sinem Kulluk, and Hülya Torun. 2017. "Training Fuzzy Neural Networks with Social Spider Algorithm." In The 5th International Fuzzy Systems Symposium (FUZZYSS'17): Programme and Abstracts, 87. Ankara, Türkiye: Fuzzy Systems Association & TOBB University of Economics and Technology. https://doi.org/10.5281/zenodo.23036889.

MLA 9: Gülmez, Burak, et al. "Training Fuzzy Neural Networks with Social Spider Algorithm." The 5th International Fuzzy Systems Symposium (FUZZYSS'17): Programme and Abstracts, Fuzzy Systems Association & TOBB University of Economics and Technology, 2017, p. 87. https://doi.org/10.5281/zenodo.23036889.

Harvard: Gülmez, B., Kulluk, S. and Torun, H. (2017) 'Training Fuzzy Neural Networks with Social Spider Algorithm', in The 5th International Fuzzy Systems Symposium (FUZZYSS'17): Programme and Abstracts. Ankara, Türkiye: Fuzzy Systems Association & TOBB University of Economics and Technology, pp. 87. doi: 10.5281/zenodo.23036889.

BibTeX:

@inproceedings{Gulmez2017training,
  author    = {Gülmez, Burak and Kulluk, Sinem and Torun, Hülya},
  title     = {Training Fuzzy Neural Networks with Social Spider Algorithm},
  booktitle = {The 5th International Fuzzy Systems Symposium (FUZZYSS'17): Programme and Abstracts},
  year      = {2017},
  month     = {oct},
  pages     = {87},
  address   = {Ankara, Türkiye},
  publisher = {Fuzzy Systems Association & TOBB University of Economics and Technology},
  doi       = {10.5281/zenodo.23036889},
  url       = {https://doi.org/10.5281/zenodo.23036889}
}