Statistics – Computation
Scientific paper
2011-04-22
Statistics
Computation
25 pages
Scientific paper
This paper studies fundamental aspects of modelling data using multivariate Watson distributions. Although these distributions are natural for modelling axially symmetric data (i.e., unit vectors where $\pm \x$ are equivalent), for high-dimensions using them can be difficult. Why so? Largely because for Watson distributions even basic tasks such as maximum-likelihood are numerically challenging. To tackle the numerical difficulties some approximations have been derived---but these are either grossly inaccurate in high-dimensions (\emph{Directional Statistics}, Mardia & Jupp. 2000) or when reasonably accurate (\emph{J. Machine Learning Research, W. & C.P., v2}, Bijral \emph{et al.}, 2007, pp. 35--42), they lack theoretical justification. We derive new approximations to the maximum-likelihood estimates; our approximations are theoretically well-defined, numerically accurate, and easy to compute. We build on our parameter estimation and discuss mixture-modelling with Watson distributions; here we uncover a hitherto unknown connection to the "diametrical clustering" algorithm of Dhillon \emph{et al.} (\emph{Bioinformatics}, 19(13), 2003, pp. 1612--1619).
Karp Dmitrii
Sra Suvrit
No associations
LandOfFree
The Multivariate Watson Distribution: Maximum-Likelihood Estimation and other Aspects does not yet have a rating. At this time, there are no reviews or comments for this scientific paper.
If you have personal experience with The Multivariate Watson Distribution: Maximum-Likelihood Estimation and other Aspects, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and The Multivariate Watson Distribution: Maximum-Likelihood Estimation and other Aspects will most certainly appreciate the feedback.
Profile ID: LFWR-SCP-O-330698