A Flexible Person Identity Authentication Archetype Based on Support Vector Machine Fusion

This paper constructs a flexible person identity authentication archetype by using multiple identity verification methods, which take effort to combine the advantages of both traditional and biometric authentication ways while overcome their respective weaknesses. By introducing fuzzification of the penalty and different cost algorithm, we can alleviate over-fitting problem and corrects the class-boundary-skew problem arising from the imbalanced training data set respectively. Besides, the overall design of the authentication strategy is presented. Finally, a flexible authentication archetype for Auto Teller Machine (ATM) is exemplified based on our strategy. Experimental results show that the SVMs with orthogonal polynomial kernels outperform GDA and those with traditional kernels in terms of generalization power and less support vectors. This archetype has versatile applications, can achieve flexible safety levels with different usability and restrain illegal access.