Extending the optimal set of discriminant vectors for an unsupervised pattern
Zhaohong Deng · Caai Transactions on Intelligent Systems · 2008
The optimal set of discriminant vectors,based on the Fisher criterion function,is an important supervised feature extraction method and has great influence in the area of pattern recognition.In this paper,an extension of the optimal set of discriminant vectors in unsupervised patterns is presented.The basic idea is to extend Fisher linear discriminants to a novel semi-fuzzy clustering algorithm through a predefined fuzzy Fisher criterion function.With the proposed algorithm,an optimal discriminant vector and fuzzy scatter matrixes can be figured out and then an unsupervised optimal set of discriminant vectors can be obtained.Experimental results for real datasets,testing clustering validity and correct classification recognition rates,demonstrated that this method is superior to the principal component analysis feature extraction algorithm in unsupervised patterns.