Enhance Neuro-fuzzy system for classification using dynamic clustering

Wongchomphu, Poonarin; Eiamkanitchat, Narissara

doi:10.1109/jictee.2014.6804071

Cited by 14 publications

(16 citation statements)

References 5 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The total average accuracy rate is higher than other high performance Neuro-fuzzy methods [5][6][7], those have been proposed earlier. The ADCNF shown the reduction of complexity and high performance than [7]. The GSS algorithm is applied to select the proper number of linguistic for rule base classification.…”

Section: Resultsmentioning

confidence: 72%

“…The experiments has done on 13 standard datasets from UCI to compare the performance of the ADCNF, ADCNF with feature selection and the ENF [7]. The 10 fold cross validation is applied to verify the performance, the result shown in table 2.…”

Section: Classification Results From Direct Calculationmentioning

confidence: 99%

“…The classification results compare with K-mean SVM, DE-SVM and DE-SVM showed a high performance. The method proposed Enhance NeuroFuzzy (ENF) in [7] is the development of [6], the algorithm to find number of membership functions using dynamic clustering is proposed. The result of classification in the method shown that the better performance.…”

Section: Introductionmentioning

confidence: 99%

“…This research focuses on the development of Neuro-fuzzy proposed in [6][7]. The concept proposed to reduce the number of membership functions.…”

Section: Introductionmentioning

confidence: 99%

“…The strength of the two algorithms is selected to use in the neuro-fuzzy system, which is the learning algorithm and the linguistic for rule extraction that is easy to understand [5][6]. The experimental in [4][5][6][7] shown that the neuro-fuzzy system has a good performance for classification. In [5][6] has proposed the neuro-fuzzy model for improving the performance of classification.…”

mentioning

confidence: 99%

See 4 more Smart Citations

The Adaptive Dynamic Clustering Neuro-Fuzzy System for Classification

Napook

Eiamkanitchat

2015

Lecture Notes in Electrical Engineering

Self Cite

View full text Add to dashboard Cite

Abstract. This paper proposes a method of neuro-fuzzy for classification using adaptive dynamic clustering. The method has three parts, the first part is to find the proper number of membership functions by using adaptive dynamic clustering and transform to binary value in a second step. The final step is classification part using neural network. Furthermore the weights from the learning process of the neural network are used as feature eliminates to perform the rule extraction. The experiments used dataset form UCI to verify the proposed methodology. The result shows the high performance of the proposed method.Keywords. Neuro-Fuzzy System, Adaptive Dynamic Clustering, Classification. IntroductionThe neural network is a popular model used for classification. It has been widely used since the strength of robustness for unknown data. Furthermore, the reason of it wildly used is the merit of using the back propagation for learning process. However, the neural network has a weak point of interpreting ability or usually called the black box [1][2][3]. The fuzzy system has been widely used in classification problems the same as the neural network. The fuzzy system can handle the uncertainty and proper for creating the linguistic value. The fuzzy rule base extraction is easy to understand and important for many applications. The weakness of the fuzzy system is no adaptive learning methodology [2]. In order to improve the classification accuracy, many researchers have proposed the combination of the neural network and fuzzy system methodology called neuro-fuzzy system. The strength of the two algorithms is selected to use in the neuro-fuzzy system, which is the learning algorithm and the linguistic for rule extraction that is easy to understand [5][6]. The experimental in [4][5][6][7] shown that the neuro-fuzzy system has a good performance for classification. In [5][6] has proposed the neuro-fuzzy model for improving the performance of classification. The method is used three fuzzy membership function for converting the original input to the linguistic value then fed into the neural network. The method applied back propagation to update neural network and fuzzy membership function. The results from training process are used to generate rule extraction for classification and can use as

show abstract