Shamanina T. Models, methods and tools for nonlinear dynamic identification of a human oculo-motor system on the basis of eye tracking technology.

The dissertation is devoted to solving the current scientific and practical problem of creating methods and means of nonlinear dynamic identification of the oculo-motor system (OMS) based on the Volterra model in the form of multidimensional transient functions (MTF) and their application in information systems for diagnosing psychophysiological state and personality recognition. Using the developed computational algorithms and software tools for processing experimental data (feedback on test visual stimuli), nonlinear dynamic models of human OMS in the form of transient functions of the first, second and third order are obtained. The obtained results of identification of OMS provide opportunities for early diagnosis of the neuro-degenerative process in Parkinson's diseases and parkinsonism syndromes, and can also be used in diagnostic studies to establish the stage of the disease; with hardware vision correction; in man-machine systems during the professional selection of operators of fast-moving technological processes, for biometric identification of users in computer systems and networks in information protection devices. A method of experimental research of human OMS has been developed using the innovative eye-tracking technology of obtaining empirical data for the identification of OMS in the form of MTF using video recording of responses to test visual stimuli with different distances from the starting position displayed on the computer monitor screen in different directions (horizontally, vertically, diagonally). Formal relations that represent universal expressions for estimating diagonal intersections of multidimensional transition functions (n-dimensional integrals from Volterra kernels) of OMS in the form of a linear combination of OMS responses to test visual stimuli have been proposed and theoretically substantiated, which made it possible to algorithmize and simplify the software implementation of the identification procedure. A method of building an approximation model of the OMS based on the Volterra polynomial using eye tracking technology is proposed. In contrast to known methods, the regularized least squares method is used to determine the diagonal intersections of multidimensional transition functions, which allows to increase the accuracy and computational stability of the identification procedure. Computational algorithms and instrumental software tools for deterministic identification of OMS based on Volterra series and polynomials in the form of MTF using test visual stimuli and eye-tracking technology have been developed. Information technology for diagnosing psychophysiological states of a person and biometric identification of a person has been developed by using as a source of primary data information models of the OMS of the «input-output» type based on Volterra series and polynomials. Eye-tracking technology is used to build models. This makes it possible to increase the accuracy of OMS modeling and, as a result, to increase the reliability of diagnosis in the space of the proposed heuristic features, which are determined using integral and differential transformations of multidimensional transition functions of OMS, which greatly simplifies the identification of features and the practical implementation of Bayesian classifiers. The results of the dissertation are implemented in the implementation of the project method in teaching students of the Small Academy of Sciences in the implementation of the innovative StartUp-project «Eye-racking in the study of cognitive processes» under the cooperation agreement between Odessa National Polytechnic University and Odessa Specialized School No. 117 The results of the dissertation are used in the educational process of the Department of Computer Systems and Software Technologies of the Odessa National Polytechnic University when teaching lecture courses, performing laboratory work and developing topics for master's theses.