Context. The problem of feature selection for big data processing based on the multi-agent approach and parallel computation has been solved. The object of research is the process of feature selection. The subject of the research are the methods of feature selection.
Objective. The purpose of the work is to create a parallel multi-agent method for reducing of big data sets.
Method. The article deals with the parallel multi-agent method for reducing of big data sets. The developed method involves splitting multiple agents into several subsets for parallel search of an informative combination of features in different areas of the search space. At the same time, it is suggested that the parallel nodes of the computer system perform the most resource-intensive operations associated with estimating the current set of agents, as well as the need to create and modify new sets of solutions based on stochastic computations. This allows to speed up the process of multi-agent search of informative combination of features, as well as to reduce the practical threshold for application of the multi-agent method with indirect communication between agents for reducing big data sets.
Results. The software which implements the proposed method and allows to select informative features based on the multi-agent approach and parallel computation has been developed.
Conclusions. The conducted experiments have confirmed the proposed software operability and allow recommending it for use in practice for solving the problems of big data processing for pattern recognition. The prospects for further research may include the modification of the developed parallel method for feature selection by using different criteria for estimation of the group information of features, as well as an experimental study of proposed method on more complex practical problems of different nature and dimensionality.

agent, data set, feature selection, parallel computing, multi-agent approach, pattern recognition