Prototype Selection from Homogeneous Subsets by a Monte Carlo Sampling
Marc Sebban · The Florida AI Research Society · 1998
In order to reduce computatiomd ald storage c~sts of learning methods, we present a protot.,~T)e selection algorithm. This apl)roach uses il,formation contained in a connected neighborhood graph. It determines the number of homogeneous subsets in the Rp space, and uses it to fix the number of prototypes in advance. Once this number is determined, we. idcntii~, prototypes applying a stratified Monte C~rlo sampling algoritl,n,. We. prc~ent an application of our algorithm oll a simulat(xl example, comparing results witil those obtained with other methods.