Cost-Effective Kernel Ridge Regression Implementation for Keystroke-Based Active Authentication System

• Published in: IEEE transactions on cybernetics Vol. 47; no. 11; pp. 3916 - 3927
• Main Authors: Pei-Yuan Wu, Chi-Chen Fang, Chang, Jien Morris, Sun-Yuan Kung
• Format: Journal Article
• Language: English
• Published: United States IEEE 01.11.2017; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Active authentication; Algorithms; Authentication; Basis functions; Biomedical monitoring; cost-effective; Feature extraction; Kernel; Kernel functions; kernel methods; kernel ridge regression (KRR); Keyboards; keystroke; Machine learning; Radial basis function; support vector machine (SVM); Support vector machines; Training; truncated-radial basis function (TRBF)
• ISSN: 2168-2267; 2168-2275
• DOI: 10.1109/TCYB.2016.2590472

In this paper, a fast kernel ridge regression (KRR) learning algorithm is adopted with O(N) training cost for largescale active authentication system. A truncated Gaussian radial basis function (TRBF) kernel is also implemented to provide better cost-performance tradeoff. The fast-KRR algorithm along with the TRBF kernel offers computational advantages over the traditional support vector machine (SVM) with Gaussian-RBF kernel while preserving the error rate performance. Experimental results validate the cost-effectiveness of the developed authentication system. In numbers, the fast-KRR learning model achieves an equal error rate (EER) of 1.39% with O(N) training time, while SVM with the RBF kernel shows an EER of 1.41% with O(N 2 ) training time.