4 층 자기연상 다층퍼셉트론을 이용한 키스트로크 기반 사용자 인증

A 4-layer AaMLP for Keystroke Dynamics Identity Verification

Password typing is the most wifely ued 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. Auto-Associative Multilayer Perceptron (AaMLP) has teen proved to be a good novelty detector. However, the wifely used 2-layer AaMLP cannot identify nonlinear boundaries, which can result in serious problems in computer security. In this paper, a nonlinear model, i.e. 4-layer AaMLP, is proposed to serve as the novelty detector, which can remedy the limitations of 2-layer AaMLP.