Investigation of the method of determination of password resistance to brute force based on an artificial neural network

Abstract

The paper deals with a topical subject of information system access security through development of a polling system aimed at forecasting the resistance of the user’s passwords against brute force. A method of development of such a system, which is based on computer-aided learning, is suggested. This method is based on the following hypothesis: “Social and psychological features of the password system users have a bearing on the complexity of created passwords.” The method is elaborated in the form of the following 6 steps: development of a group of questions which reflect the preferences of the password system user; implementation of a polling system which requests the user’s typical password; selection of a mechanism of independent estimation of password complexity; application of the polling system to generate implicit interrelations of specific features of users and their passwords; development and training of a neural network to solve a regression problem based on the results of the polling system use; and estimation of the neural network by a test set of users. The experiment results are presented in a form of a diagram. The results of the experiment are discussed and conclusions are drawn about the method development.