MODELING CONDITIONAL-PROBABILITY DISTRIBUTIONS FOR PERIODIC VARIABLES

Authors
BISHOP CM NABNEY IT
Citation
Cm. Bishop et It. Nabney, MODELING CONDITIONAL-PROBABILITY DISTRIBUTIONS FOR PERIODIC VARIABLES, Neural computation, 8(5), 1996, pp. 1123-1133
Citations number
13
Categorie Soggetti
Computer Sciences","Computer Science Artificial Intelligence",Neurosciences
Journal title
Neural computation → ACNP
ISSN journal
08997667
Volume
8
Issue
5
Year of publication
1996
Pages
1123 - 1133
Database
ISI
SICI code
0899-7667(1996)8:5<1123:MCDFPV>2.0.ZU;2-F
Abstract
Most conventional techniques for estimating conditional probability de nsities are inappropriate for applications involving periodic variable s. In this paper we introduce three related techniques for tackling su ch problems, and investigate their performance using synthetic data. W e then apply these techniques to the problem of extracting the distrib ution of wind vector directions from radar scatterometer data gathered by a remote-sensing satellite.