Identifying User Authentication and Most Frequently Used Region Based on Mouse Movement Data: A Machine Learning Approach

User authentication in computer demands a reliable, highly secure and fast system so that it can address many severe attacks, which are unable to handle by the conventional password system. Several behavioral biometrics has already progressed to overcome these problems. Mouse-operating behavior of a user is prevalent as powerful behavioral biometrics for access control. It requires detecting authorized and unauthorized users within a very short period of time while a user interacts initially with the computer. In this correspondence paper, we present the mouse dynamics biometric authentication approach that shows the performance from five and ten second's data of mouse movements. The features are extracted from the given dataset to find out the most frequently used regions. In order to identify the accuracy, the extracted features are evaluated and accomplish accuracy using 5 machine learning classification algorithms. The extensive experimental result proves, random forest algorithm outperforms than others and the efficacy of the proposed method is evaluated by False Positive Rate achieves 1.74%, False Negative Rate of 29.23% and 87.23% classification accuracy. An additional experiment is carried to compare the notable improvement of the proposed approach with other methods in the literature.