Aggregation is the conjunction of information, aimed at its compact represen tation. Any time when the totality of data is described in terms of general ized indicators, conventional counts, typical representatives and characteristic dependences, one directly or indirectly deals with aggregation. It includes revealing the most significant characteristics and distinctive features, quanti tative and qualitative analysis. As a result, the information becomes adaptable for further processing and convenient for human perception. Aggregation is widely used in economics, statistics, management, planning, system analysis, and many other fields. That is why aggregation is so important in data pro cessing. Aggregation of preferences is a particular case of the general problem of ag gregation. It arises in multicriteria decision-making and collective choice, when a set of alternatives has to be ordered with respect to contradicting criteria, or various individual opinions. However, in spite of apparent similarity the problems of multicriteria decision-making and collective choice are somewhat different. Indeed, an improvement in some specifications at the cost of worsen ing others is not the same as the satisfaction of interests of some individuals to the prejudice of the rest. In the former case the reciprocal compensations are considered within a certain entirety; in the latter we infringe upon the rights of independent individuals. Moreover, in multicriteria decision-making one usu ally takes into account objective factors, whereas in collective choice one has to compare subjective opinions which cannot be measured properly.

This monograph presents the author's studies in music recognition aimed at developing a computer system for automatic notation of performed music. The performance of such a system is supposed to be similar to that of speech recognition systems: acoustical data at the input and music scoreprinting at the output. The approach to pattern recognition employed is thatof artificial perception, based on self-organizing input data in order to segregate patterns before their identification by artificial intelligencemethods. The special merit of the approach is that it finds optimal representations of data instead of directly recognizing patterns.