Nonparametric Algorithms and Classification Systems

摘要

The techniques of synthesis and analysis of nonparametric algorithms and classification systems based on Rosenblatt-Parzen estimates of probability density and sequential decision-making procedures are studied. Special attention is paid to the classification problems in the context of large and small samples, to pattern recognition in the circumstances of ambiguous "teacher instructions," and to the minimization of description. The results are used for statistical modeling of developing systems and the construction of an algorithm searching for a global extremum.