THE MIVAR APPROACH AS A BASIS FOR A QUALITATIVE TRANSITION TO A NEW LEVEL IN THE FIELD OF ARTIFICIAL INTELLIGENCE

摘要

The mivar approach main originality is that three mivar technologies are combined on the basis of epistemological multidimensional evolutionary databases and the «Thing–Property–Relation» rules in which the automatic algorithm constructor with linear computational complexity and/or the logical inference on bipartite directed mivar networks «Object–Rules» are implemented. Mivar practical using proved that logical inference is made now on an ordinary laptop with a speed of more than 5 million product rules / second. The task of logical inference on products must be excluded from the class of NP-complete tasks, because mivar method of logical inference with a linear computational complexity solution not using either a full search of options or heuristics is found and patented. In computer science mivars solve the problems of accumulating and processing information. By results of 2017 software products based on the miware approach have already been created for: Expert systems – Wi!Mi; Razumator; Texts meaning understand – Tel!Mi; Decision-making systems for autonomous robots – Robo!Razum. Wi!Mi technology is used in Tel!Mi for meaning understanding. Tel!Mi automatically allows to teach Wi!Mi Razumator and Robo!Razum with the texts of the instructions. In total these products can be used for the Internet of things and other cyberphysical systems since mivar products Wi!Mi, Tel!Mi, Robo!Razum can work on one usual (rather cheap) computing module «processor-memory». So the mivar approach to the accumulation of data and information processing combines the main directions of artificial intelligence at the logical level of research: expert systems, language understanding, image recognition, ACS and intelligent autonomous robots. Therefore it is the mivar approach that is the basis for a qualitative transition to a new level in the artificial intelligence field.