On the improvement of brain tumour data clustering using class information
Raúl Cruz, Alfredo Vellido · Starting AI Researchers' Symposium · 2006
Cluster analysis can benefit from the use of class labels, if these are available, in a semi-supervised approach. In this study, we propose the integration of class information in the clustering of Magnetic Resonance Spectra (MRS) corresponding to human brain tumours using an extension of Generative Topographic Mapping (GTM) that behaves robustly in the presence of outliers.