Training procedures which ensure the job reliability in pattern recognition

Research object: recognition of technical and geological objects states. The paper aims to develop theoretical basis and principles of constructing efficient training procedures ensuring the higher quality and reliability of recognition. New training algorythms have been developed to ensure the higher quality and reliability of recognition. The automatic recognizing systems have been constructed. Their quality and reliability in recognition is not less than the needed levels. The algorythms and application package of 'Puma-Priznak' have been introducedAvailable from VNTIC / VNTIC - Scientific & Technical Information Centre of RussiaSIGLERURussian Federatio