In this correspondence, we propose a novel class of learning vector qu
antizers (LVQ's) based on multivariate data ordering principles. A spe
cial cast of the navel LVQ class is the median LVQ, which uses either
the marginal median or the vector median as a multivariate estimator o
f location. The performance of the proposed marginal median LVQ in col
or image quantization is demonstrated by experiments.