29-38
Search Fuzzy Image of the Attacker Based on the Use of Automatic Classification Methods
Authors: Simon Zh. Simavoryan, Arsen R. Simonyan, Elena I. Ulitina, Rafik A. Simonyan, Ellina A. Pilosyan, Nadezhda A. Kornienko
Number of views: 458
Work is devoted to development of methods of search of an indistinct image of the malefactor on the basis of use of methods of automatic classification. This type of search is used by initial search when the inquiry is set in the form of the description of signs of malicious action. Search is carried out on data from the specialized knowledge base on malicious actions (templates) and the knowledge base on regular situations and values of their admissible characteristics. Methods of automatic classification meet as well other names: objective classification, splitting, taxonomy, diagonalization of matrixes of communication, etc. When using these methods, given about malicious actions, join in the knowledge base as an image (n + 1) member malicious action. After that some algorithm of automatic classification which is carrying out splitting the knowledge base into groups of images of malicious actions "uniform" somewhat is put in action. As result of initial search of an image of the malefactor system the images of malefactors carried by an algorithm of automatic classification in the same group, as an image of the malefactor for whom the image of malicious action is looked for are given. For increase in a noise stability of the procedure of initial search of an indistinct image of the malefactor it is possible to carry out in parallel splitting on various algorithms of automatic classification then as an indistinct image of the malefactor to give association (for increase in completeness of initial search), crossing (for completeness of accuracy), or composition of subsets in which has been carried (n + 1) member an image from the knowledge base.