Report ID
1995-20
Report Authors
Omer Egecioglu and Ashok Srinivasan
Report Date
Abstract
Non-parametric density estimation is the problem of approximating the values ofa probability density function, given samples from the associateddistribution. Non-parametric estimation finds applications in discriminantanalysis, cluster analysis, and flow calculations based on Smoothed ParticleHydrodynamics. Usual estimators make use of kernel functions, and require onthe order of $n^2$ arithmetic operations to evaluate the density at $n$ samplepoints. We describe a sequence of special weight functions which requiresalmost linear number of operations in $n$ for the same computation.
Document
1995-20.ps212.7 KB