Формализация «естественной» классификации и систематики через неподвижные точки предсказаний

Nowadays there exist many approaches to classification and clustering; for instance one can mention those based on compactness and various metrics on feature spaces, based on etalons, on distributions composition partitioning, etc. In contrast to these approaches, the task of “natural” classification is to discover a classification as a law of nature that satisfy some requirements promoted by naturalists. The sense of this law is in the compression of information by extracting the structure of natural objects. We propose a formalization of this law based on fix-points of probabilistic laws of special type. We prove that the probabilistic laws we define solve the problem of statistical ambiguity and thus they enable us to predict without contradictions and to provide consistent fix-points. These fix-points form a “natural” classification. Finally we present the results of a computer experiment on building and recognition of classes of transcription factors binding sites.