Breast Cancer Diagnosis via Fuzzy Clustering with Partial Supervision

Clustering is a procedure in which objects are distiguished or classified in accordance with their similarity. There is no teacher to provide guidance, hence it is also called unsupervised classification. According to the theory of classification, clustering methods may be treated as classification methods that utilize minimal information about classified objects (their features). A data set partition can be described by a c×N partition matrix U (where c is the number of clusters, N is the number of objects) [1], where each element of U represents the membership of every input object in fuzzy clusters. Clustering results can either be used, after hardening of the partition matrix, as a final partition of the input data or be processed further by a human expert, expert systems and so on.