Password is the most widely used identity verification method in computer security domain. However, due to its simplicity, it is vulnerable to imposter attacks. Keystroke dynamics adds a shield to password. Discriminating imposters from owners is a novelty detection problem. Recent research reported good performance of Auto-Associative Multilayer Perceptron(AaMLP). However, the 2-layer AaMLP cannot identify nonlinear boundaries, which can result in serious problems in computer security. In this paper, we applied 4-layer AaMLP as well as SVM as novelty detector to keystroke dynamics identify verification, and found that they can significantly improve the performance.
展开▼