Skobeltsyn Institute of Nuclear Physics
Lomonosov Moscow State University

The subject of the paper is an application of the pattern recognition theory based on multidimensional statistical approach to solving the classification problem appearing in cosmic rays physics. The paper fully describes the algorithm of the primary cosmic particles separation into 2 classes, developed on basis of the Bayesian linear classi-fier. As an illustration the practical computational problem of the primary particles separation by location of their first interaction point within the device (spectrometer of the NUCLEON scientific project) is solved with the aid of the suggested algorithm using model data. As distinct from the primary energy determination problem in the above-mentioned NUCLEON project, the practical computational problem under con-sideration cannot be solved by any method, even approximate one, within the bounds of the deterministic (not statistical) approach.
In addition, a wide range of the experimental cosmophysics problems to be solved by the technique suggested is briefly outlined.