TY - GEN
T1 - Optimized FCM-based radial basis function neural networks
T2 - 7th International Symposium on Neural Networks, ISNN 2010
AU - Kim, Wook Dong
AU - Oh, Sung Kwun
AU - Huang, Wei
PY - 2010
Y1 - 2010
N2 - In this paper, we introduce a new architecture of optimized FCM-based Radial Basis function Neural Network by using space search algorithm and discuss its comprehensive design methodology. As the consequent part of rules of the FCM-based RBFNN model, four types of polynomials are considered. The performance of the FCM-based RBFNN model is affected by some parameters such as the number of cluster and the fuzzification coefficient of the fuzzy clustering (FCM) and the order of polynomial standing in the consequent part of rules, we are required to carry out parametric optimization of network. The space evolutionary algorithm(SEA) being applied to each receptive fields of FCM-based RBFNN leads to the selection of preferred receptive fields with specific local characteristics available within the FCM-based RBFNN. The performance of the proposed model and the comparative analysis between WLSE and LSE are illustrated with by using two kinds of representative numerical dataset.
AB - In this paper, we introduce a new architecture of optimized FCM-based Radial Basis function Neural Network by using space search algorithm and discuss its comprehensive design methodology. As the consequent part of rules of the FCM-based RBFNN model, four types of polynomials are considered. The performance of the FCM-based RBFNN model is affected by some parameters such as the number of cluster and the fuzzification coefficient of the fuzzy clustering (FCM) and the order of polynomial standing in the consequent part of rules, we are required to carry out parametric optimization of network. The space evolutionary algorithm(SEA) being applied to each receptive fields of FCM-based RBFNN leads to the selection of preferred receptive fields with specific local characteristics available within the FCM-based RBFNN. The performance of the proposed model and the comparative analysis between WLSE and LSE are illustrated with by using two kinds of representative numerical dataset.
KW - Fuzzy C-means clustering
KW - Machine learning data
KW - Radial basis function neural network
KW - Space evolutionary algorithm
UR - http://www.scopus.com/inward/record.url?scp=77954453331&partnerID=8YFLogxK
U2 - 10.1007/978-3-642-13278-0_27
DO - 10.1007/978-3-642-13278-0_27
M3 - Conference contribution
AN - SCOPUS:77954453331
SN - 3642132774
SN - 9783642132773
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 207
EP - 214
BT - Advances in Neural Networks - ISNN 2010 - 7th International Symposium on Neural Networks, ISNN 2010, Proceedings
Y2 - 6 June 2010 through 9 June 2010
ER -