Abstract:
We consider the classification problem when the sample size and the number of dimensions of random variables are growing simultaneously. The dependence of classification errors on the estimates of the first and the second moments of discrimination function and on the ratio of the number of estimated parameters to the sample size is investigated.