There are grounds to consider that pattern recognition in geological sciences is a challenge to those engaged in applied mathematics, which was cast in the 50-s of the XX century by the practice of scientific investigations, to which no worth answer was found . As it seems, this is caused to a considerable extent by the fact that schemes and norms of reasonings and algorithms of applied mathematics which turned out to be not much appropriate outside theoretical physics, are amployed in pattern recognition.
The application of the pattern recognition results indicates to the fact that up to now there has been no ansver to such questions as: when pattern recognition has scientific sense and when not?, when such a sense can be given to pattern recognition and how to do this without changing the system of concepts and the scheme of data acquisition?
One can assume that the pattern recognition is, in a sense, not a construction of an optimal algorithm, but a construction due to accumulation of empirical data, of an algorithms sequence, each subsequent one being more preferable than the previous one. In other words, the pattern recognition is, first of all, the preference of pattern recognition algorithms.
The pattern recognition does not imply immediate substitution of special automated system, but their subsequent and substantiated ousting.
The pattern recognition should be finished with the evident evaluation of the advantages to be gained from the use of expensive and sophisticated automated systems instead of the conventional speculations as well as the estimation of losses due to a dishonest advertising of these automated systems.