Title :
Frequency sensitive competitive learning for clustering on high-dimensional hyperspheres
Author :
Banerjee, Arindam ; Ghosh, Joydeep
Author_Institution :
Dept. of Electr. & Comput. Eng., Texas Univ., Austin, TX, USA
fDate :
6/24/1905 12:00:00 AM
Abstract :
This paper derives three competitive learning mechanisms from first principles to obtain clusters of comparable sizes when both inputs and representatives are normalized. These mechanisms are very effective in achieving balanced grouping of inputs in high dimensional spaces as illustrated by experimental results on clustering two popular text data sets in 26,099 and 21,839 dimensional spaces, respectively
Keywords :
data handling; maximum likelihood estimation; neural nets; pattern clustering; unsupervised learning; balanced grouping; competitive learning; frequency sensitive competitive algorithm; high dimensional text data sets; high-dimensional hypersphere; maximum likelihood estimation; spherical k-means algorithm; text clustering; winner take-all networks; Clustering algorithms; Euclidean distance; Frequency; Hebbian theory; Learning systems; Power capacitors; Resource management; Stability; Subspace constraints; Vector quantization;
Conference_Titel :
Neural Networks, 2002. IJCNN '02. Proceedings of the 2002 International Joint Conference on
Conference_Location :
Honolulu, HI
Print_ISBN :
0-7803-7278-6
DOI :
10.1109/IJCNN.2002.1007755
