Autors: Chandramohan,V., Pham,T.D.
Title: Cancer Classi cation using Kernelized Fuzzy C-means
Keywords: Mass spectrometry, feature extraction, kernel fucntions, classi cation

Abstract: Pattern analysis of mass spectra obtained from blood samples, has attracted the attention for early detection of cancer. In this paper, we present an unsupervised kernel based fuzzy c-means algorithm(KFCM), which is realized by modifying original Euclidean distance in classical fuzzy clustering algorithm(FCM) by kernel-induced distance metric. Our analysis on mass spectrometry dataset, shows that KFCM has better clustering performance and is more robust to noise than FCM. We evaluated the performance of our classi cation methods with some popular classi cation techniques like SVM, PCA, LDA/QDA and randomforests.

References

    Issue

    9th WSEAS International Conference on FUZZY SYSTEMS (FS’08), pp. 90-99, 2008, Bulgaria,

    Вид: публикация в национален форум