Summary: | The paper presents elements of the theory of collective intelligence systems based on information technology of evolutionary coordination of solutions. The teamwork coordinator is genetic algorithms. A review of the current state of the theory and practice of collective intelligence systems is given and it is concluded that the creation and development of such a theory is necessary to reduce the likelihood of errors in solving difficult problems. The evolutionary matching method proposed and developed by the authors is considered, which has the properties of increasing the probability of making the right decisions of problems of medium difficulty compared to the best actor in the group and significantly reducing the probability of erroneous decisions in difficult cases. The corresponding theorems are formulated and proved for this method. The results of computer simulations confirming these effects are presented, as well as the results of a drastic decrease in the probability of errors of the first kind in image recognition using neural network committees are presented.
|