TY - GEN
T1 - Feature subset-wise mixture model-based clustering via local search algorithm
AU - Namkoong, Younghwan
AU - Joo, Yongsung
AU - Dankel, Douglas D.
PY - 2010
Y1 - 2010
N2 - In clustering, most feature selection approaches account for all the features of the data to identify a single common feature subset contributing to the discovery of the interesting clusters. However, many data can comprise multiple feature subsets, where each feature subset corresponds to the meaningful clusters differently. In this paper, we attempt to reveal a feature partition consisting of multiple non-overlapped feature blocks that each one fits a finite mixture model. To find the desired feature partition, we used a local search algorithm based on a Simulated Annealing technique. During the process of searching for the optimal feature partition, reutilization of the previous estimation results has been adopted to reduce computational cost.
AB - In clustering, most feature selection approaches account for all the features of the data to identify a single common feature subset contributing to the discovery of the interesting clusters. However, many data can comprise multiple feature subsets, where each feature subset corresponds to the meaningful clusters differently. In this paper, we attempt to reveal a feature partition consisting of multiple non-overlapped feature blocks that each one fits a finite mixture model. To find the desired feature partition, we used a local search algorithm based on a Simulated Annealing technique. During the process of searching for the optimal feature partition, reutilization of the previous estimation results has been adopted to reduce computational cost.
KW - Clustering
KW - Feature selection
KW - Finite mixture model
UR - http://www.scopus.com/inward/record.url?scp=77953729493&partnerID=8YFLogxK
U2 - 10.1007/978-3-642-13059-5_15
DO - 10.1007/978-3-642-13059-5_15
M3 - Conference contribution
AN - SCOPUS:77953729493
SN - 3642130585
SN - 9783642130588
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 135
EP - 146
BT - Advances in Artificial Intelligence - 23rd Canadian Conference on Artificial Intelligence, Canadian AI 2010, Proceedings
T2 - 23rd Canadian Conference on Artificial Intelligence, Canadian AI 2010
Y2 - 31 May 2010 through 2 June 2010
ER -