TY - GEN
T1 - Ultrametricity of dissimilarity spaces and its significance for data mining
AU - Simovici, Dan
AU - Vetro, Rosanne
AU - Hua, Kaixun
PY - 2015
Y1 - 2015
N2 - We introduce a measure of ultrametricity for dissimilarity spaces and examine transformations of dissimilarities that impact this measure. Then, we study the influence of ultrametricity on the behavior of two classes of data mining algorithms (kNN classification and PAM clustering) applied on dissimilarity spaces. We show that there is an inverse variation between ultrametricity and performance of classifiers. For clustering, increased ultrametricity generate clusterings with better separation. Lowering ultrametricity produce more compact clusters.
AB - We introduce a measure of ultrametricity for dissimilarity spaces and examine transformations of dissimilarities that impact this measure. Then, we study the influence of ultrametricity on the behavior of two classes of data mining algorithms (kNN classification and PAM clustering) applied on dissimilarity spaces. We show that there is an inverse variation between ultrametricity and performance of classifiers. For clustering, increased ultrametricity generate clusterings with better separation. Lowering ultrametricity produce more compact clusters.
UR - https://www.scopus.com/pages/publications/84994666080
UR - https://www.scopus.com/pages/publications/84994666080#tab=citedBy
M3 - Conference contribution
AN - SCOPUS:84994666080
T3 - Revue des Nouvelles Technologies de l'Information
SP - 89
EP - 100
BT - Extraction et Gestion des Connaissances, EGC 2015
PB - Hermann-Editions
T2 - Quinzieme Conference Internationale Francophone sur l'Extraction et la Gestion des Connaissances, EGC 2015 - 15th International French-Speaking Conference on Knowledge Extraction and Management, EGC 2015
Y2 - 27 January 2015 through 30 January 2015
ER -