Keystroke Authentication Using One-Class Classifier

The patent extracts and removes the requirement for multiple users to type the same content from the authentication process. Instead of using two-class classifiers that need training data from multiple users, the system uses one-class classifiers that only require data from a single user, thereby eliminating the training complexity and security risks associated with collecting multiple users' typing data while maintaining authentication accuracy

The patent introduces an intermediary approach by using one-class classifiers as a middle ground between simple distance-based methods and complex two-class classifiers. This intermediary solution maintains the accuracy benefits of two-class classifiers while avoiding their requirement for multiple users' training data, thus resolving the contradiction between accuracy and training complexity

The patent removes the security vulnerability inherent in two-class classifier methods by eliminating the requirement to collect and store multiple users' typing data on sensitive information. By using one-class classifiers that only need a single user's enrollment data, the system maintains authentication accuracy while eliminating the security risks associated with handling multiple users' sensitive information

The patent converts the potential harm of requiring multiple users to type sensitive information into a benefit by using one-class classifiers. The system achieves comparable or better authentication accuracy without the security risks, effectively turning the limitation into an advantage by reducing the attack surface and data protection requirements