Resource development to prevent riots at mass events

Currently, the main communication means and management of the participants behavior in mass events, including unauthorized ones, are portable mobile devices, through which, in the social network framework a, text messages are sent to manipulate people's behavior. In the article the problem of mass events participants binary classification on representing and not representing a threat is considered. A convolutional neural network is used as a classification tool. The system of loading and processing data from a social network, the neural network architecture a and the results of experiments are described.