Intruder Detection Method By Evaluating The Input -New Approach To GUI System-

摘要

In this research, we focused on the characteristics that some people how use the mouse or keyboard when they using interactive software. Especially when people use a mouse device, how do they move on different from other people, such as coordinates, trajectory, speed and movement clicking speed. We saved up mouse clicking's log data and used it to extract features of the individual.We propose a new intruder detection method that used mouse clicking's log data.In this intruder detection method, we evaluate the input command sequence for authentication using the transition probability table. This method has two steps. First step, we divided the collected into two sets , as the data in the first half of the training data, and used it to created a transition probability table. Next step, make the second half data into validation data, and used the transition probability table to take the average of the total. We appropriate the average value's height to judge who is real user.This method is easy to implement to an ordinary computer_ Using this method in conjunction with traditional authentication methods such as password system. And it can help system security be stronger than before.