Title :
Multiple source clustering: a probabilistic reasoning approach
Author :
Leih, Thomas J. ; Harmse, Jørgen ; Giannopoulos, Evangelos
Author_Institution :
Tracor Appl. Sci., Austin, TX, USA
Abstract :
In this paper we describe a versatile multiple source clustering (MSC) algorithm. The algorithm uses a form of probabilistic reasoning known as Bayesian networks to solve the MSC problem of incomparable feature spaces. For time-tagged data, the algorithm uses fuzzy conjunctions to support cluster formation and management. Clustering performance measures are defined and a multiple target tracking/multiple sensor example is presented
Keywords :
Bayes methods; belief maintenance; fuzzy logic; inference mechanisms; probability; sensor fusion; target tracking; tracking; Bayesian networks; beliefs; clustering performance measures; feature extraction; fuzzy logic; incomparable feature spaces; multiple sensor; multiple source clustering; multiple target tracking; probabilistic reasoning; time-tagged data; Bayesian methods; Clustering algorithms; Computer networks; Data mining; Extraterrestrial measurements; Q measurement; Target tracking; Uncertainty;
Conference_Titel :
Data Fusion Symposium, 1996. ADFS '96., First Australian
Conference_Location :
Adelaide, SA
Print_ISBN :
0-7803-3601-1
DOI :
10.1109/ADFS.1996.581097
