PG-means: Learning the Number of Clusters in Data

منو

" PG-means: Learning the Number of Clusters in Data "

Document Type	:	Latin Dissertation
Language of Document	:	English
Record Number	:	149164
Doc. No	:	ET20956
Main Entry	:	Yu Feng
Title Proper	:	PG-means: Learning the Number of Clusters in Data
Note	:	This document is digital این مدرک بصورت الکترونیکی می باشد
Abstract	:	We present a novel algorithm called PG-means in this thesis. This algorithmis able to determine the number of clusters in a classical Gaussian mixture modelautomatically. PG-means uses efficient statistical hypothesis tests on one-dimensionalprojections of the data and model to determine if the examples are well representedby the model. In so doing, we apply a statistical test to the entire model a t once,not just on a per-cluster basis. We show that this method works well in difficultcases such as overlapping clusters, eccentric clusters and high dimensional clusters.PG-means also works well on non-Gaussian clusters and many true clusters. Further.the new approach provides a.
Subject	:	Electericl tess
	:	برق
electronic file name	:	TL44102.pdf
Title and statement of responsibility and	:	PG-means: Learning the Number of Clusters in Data [Thesis]