Open AccessQuantitative methods of classification of vegetation: condition, problems, optimization
Author(s) -
G. Rozenberg
Publication year - 2020
Publication title -
sbornik naučnyh trudov
Language(s) - English
Resource type - Journals
ISSN - 0201-7997
DOI - 10.36305/2712-7788-2020-1-154-45-55
Subject(s) - dendrogram , vegetation (pathology) , similarity (geometry) , pattern recognition (psychology) , computer science , reduction (mathematics) , artificial intelligence , simple (philosophy) , data mining , mathematics , image (mathematics) , medicine , population , philosophy , demography , geometry , epistemology , pathology , sociology , genetic diversity
The main stages of creating automatic vegetation classification procedures (contingency indices, similarity coefficients, simple algorithms for automatic classification - construction of dendrograms, dendrites, correlation pleiades, etc.) are discussed. The reduction of the number of features (types) is considered as the first condition for the optimization of the classification procedure. The results of the experiment on the reduction of species with the automatic classification of 50 descriptions of Achnatherum splendens (Trin.) Nevski formation in the floodplain of theTuulRiverinMongoliaare discussed. It is concluded that new successes in the automatic classification of vegetation should be expected not in the direction of developing some new methods, but in advancing new ideas about the structure and character of the dynamics of plant communities (paradigm shift).