A novel algorithm to model the neuromuscular system from the eye to fingers to authenticate individuals through a typing process

The extensive use of computers has necessitated a new paradigm, in which computers are not only the major channel for dealing with day-to-day financial, industrial, and individual duties, but also the need to establish effective user identification for authentication reasons. Based on this fact, behavioral biometrics, such as typing, can be used for authentication to be subtle, unlike most biometrics. In this paper, to verify the identity, an adaptive neuro-fuzzy inference system (ANFIS) is employed to model musculoskeletal system from the eye to the fingers in the typing process, as well as to model the control process of typing behavior. Model predictive control (MPC) is used to model the control process in order to get the best results. The improved distance evaluation (IDE) feature selection technique is utilized to minimize feature dimensions, and data fusion is conducted at the feature level. Besides, the Support Vector Machine (SVM) classifier is applied to authenticate selected features. Moreover, this algorithm is tested on a dataset of 35 users, providing accuracy with an Arithmetic mean of 99.65.