Proposal of Rhythm Authentication Method Using Users Classification by Self-Organizing Map

The authentication methods, e.g. PINs and patterns, are used on screen lock for mobile devices. These methods are vulnerable to attacks by peeking. This problem is hard to solve, as long as the user input the information of authentication while he/she watches the screen. In this paper, we propose the rhythm authentication method that uses a user classification by the self-organizing map (SOM) using the event times and coordinates of tapping point on a screen. This method is not peeked by others, because the user can tap the screen only, and not needs to watch the screen. Therefore, we expect to solve the shoulder-surfing attack vulnerability by our approach.