004.82The problem of creation of a new class of computers called pattern computers is considered. These computer systems are based on the concept of representation of information on their environment in the form of patterns. These computers also process data represented in the form of patterns. Patterns correspond to objects of the world, can support data interchange with one another, and can interact with one another, whereas pattern computers model these activities. A version of description of patterns and examples of representations of patterns are proposed.The creation of high-speed computers with large memory and perfect software became one of the most substantial technical achievements of the twentieth century.Computers that excellently solve isolated specialized problems, for example, successfully play chess even with Grand Masters, are faced with difficulties in solving intellectual problems that are characteristic of man. But it is precisely the latter problems that are of greatest interest to the user who would like to use a computer as a clever partner and not only as a device for execution of previously written and debugged programs.Why computers are so inefficient in solving problems that are easily solved not only by man but also animals? For example, a dog recognizes the exterior of his master, distinguishes him from other persons, obeys his orders, and performs these functions in real time, whereas the solution of a similar class of problems is a very complicated problem for computers. Computers are not yet focused on the solution of problems that are connected with processing large bodies of interrelated nonhomogeneous data.From the author's viewpoint, the problem is that the solution of problems of recognition, identification, understanding of meaning, construction of logical reasoning chains, simulation of external environments, creation of hypotheses, and generation of strategies of behavior requires the processing of large bodies of data whose forms and means of representation are different. Of course, general methods exist that lie in the construction of a special language that allows one to formalize a problem in a certain manner in order to solve it on a computer or in the construction of specialized computers. But the construction of a special language for solving a class of problems makes no sense since the complexity of such a language is comparable with the complexity of solving problems from this class.Moreover, a large body of data can be stored in computers, but then the search for and choice of data items that are necessary for solution of a formulated problem becomes a problem. A way out of this difficulty is a data structuring tool such that, first, it makes it possible to use each found element as a means of reduction in overall search. Second, it begins to process the found data items before the completion of the entire search. Third, it chooses the processing methods that are best suited to the data items related to the problem being considered.It is a probable way of solution...