Sample dispersion is better than sample discrepancy for classification

We want to generate learning data within the context of active learning. First, we recall theoretical results proposing discrepancy as a criterion for generating sample in regression. We show surprisingly that theoretical results about low discrepancy sequences in regression problems are not adequate for classification problems. Secondly we propose dispersion as a criterion for generating data. Then, we present numerical experiments which have a good degree of adequacy with theory.