Collective of many-dimensional nonparametric regressions, grounded on decomposition of learning sample according to its size

The technique of synthesis and analysis of collective of many-dimensional nonparametric regressions, which provides for high computing efficiency of problem solving of restoration of stochastic dependencies, at the account of usage of technology of parallel calculations, is offered. Asymptotic properties of collective are researched, results of their comparison with properties of traditional nonparametric regression, are presented.