A Typing Pattern Authentication Mechanism Based on Radial Basis Functions Neural Networks

Information security has become one of major fields in present time. Millions of dollars losses have motivated engineers to rethink on security measures. In this way efficient authentication and identification of users guarantees computer systems protection from unauthorized access. In this paper, keystroke or keyboard dynamics biometric patterns are selected to reinforce a password-based authentication system. A radial basis function neural network (hereafter RBF) with various number of neurons in its hidden layer is utilized to act as a keystroke patterns classifier. Also the optimality of RBF neural network with a specified number of the hidden layer neurons is evaluated according to equal error rate (EER). In addition it is demonstrated that RBF neural networks can be trained faster than traditional feed-forward neural networks like radial basis function (RBF).

کلیدواژه ها:

Radial basis function neural network (RBFNN) ، Network security ، Keystroke dynamics ،