Objective evaluation of knitted yarn quality based on fuzzy kernel C-means
- 1 June 2010
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- Vol. 2, 45-48
Abstract
In this paper, the method that measuring dataset of knitted yarns is clustered using improving fuzzy kernel c-Means (FKCM) clustering algorithm is proposed. In FKCM clustering algorithm, the data of low dimension input space is mapped to high dimension feature space, FCM clustering algorithm is performed in feature space, then the constraint optimization distance matrix and membership matrix of testing samples are computed by utilizing iterative algorithm, and the clustering result can be acquired according maximum membership principle. Subsequently, the Kernel F cluster validity index is designed for seeking the fitness cluster number and the corresponding relationship model of clusters sequence number and quality grades is constructed. Improving FKCM clustering algorithm is more efficient than other clustering algorithms, and clustering result can provide the training samples for constructing quality grades and clusters recognition function of new samples. The combination of improving FKCM and KF index provides an efficient data analysis method for multi-index dataset.Keywords
This publication has 3 references indexed in Scilit:
- Mercer kernel-based clustering in feature spaceIEEE Transactions on Neural Networks, 2002
- Input space versus feature space in kernel-based methodsIEEE Transactions on Neural Networks, 1999
- Convergence theory for fuzzy c-means: Counterexamples and repairsIEEE Transactions on Systems, Man, and Cybernetics, 1987