A Framework for Improving the Accuracy of Keystroke Dynamics-Based Biometric Authentication Using Soft Computing Techniques

The global access of information and resources from anywhere has increased the chance of intrusion and hacking of confidential data. Username with password is the commonly used authentication mechanism which is used for almost all online applications from net banking to online examinations. However, advanced safeguard mechanisms are sought against unauthorized access as the passwords can be hacked easily. To strengthen the password, it can be combined with biometric technology. Keystroke biometrics, a strong behavioral biometric, can be considered as a secure method compared to other methods even if the imposter knows the username and password as it is based on user habitual typing rhythm patterns. For any online application the accuracy level plays a vital role. But the accuracy of keystroke authentication when compared with other biometric authentication mechanisms is low. To improve the accuracy and minimize the training and testing time, this chapter proposes a wrapper-based classification using PSO-ELM-ANP algorithm which gives 99.92% accuracy.