COMPLEXITY AND LEVEL LOGICAL DESCRIPTION OF CLASSES FOR PATTERN
RECOGNITION PROBLEMS
In the frameworks of logic-axiomatical approach to the pattern recognition
problem the guaranteed time effectiveness of the solving algorithms is proved
by means of consideration of several variants of setting of a problem
including the problem of analysis of a compound object.
By the author a level description of classes was suggested. Such a description allows to
decrease upper bounds of step number of algorithms solving pattern recognition
problems. In this paper conditions for such a decreasing are proved.