Independent Study Presentation

CS7387 RESEARCH IN COMPUTER SCIENCE

Title: A study on the generalizability of Oculomotor Plant Mathematical Model

Presenter:  Dmytro Katrychuk

Advisor:  Dr. Oleg Komogortsev

Date/Time:  Thursday, December 2 @3:00 – 3:30 p.m.

Location:  ZOOM:  https://txstate.zoom.us/j/91243653692

Abstract:

The Oculomotor plant mathematical model (OPMM) is a dynamic system that describes a human eye in motion. In this study, we focus on an anatomically inspired model, every part of which represents a certain biological component of a real oculomotor plant. As a first step, we consider a model that is capable of reproducing eye gaze trajectories during horizontal saccade – a rapid movement of an eye from one fixation point to another. In this simplified case, the system consists of two muscles attached to the eyeball: agonist (one that pulls toward the movement) and antagonist (one that opposes it), both of which are modeled as a set of linear springs. The parameters of the model, therefore, correspond to the properties of the eyeball, muscles, and a neuronal control signal that stems from the brain.

An anatomically accurate eye model can capture an internal state of the oculomotor plant given its observed behavior in the form of eye movements. In the past, the utility of such models was proven to be promising in biometrics and eye movement prediction frameworks. A shared underlying assumption is that a set of parameters estimated for a certain subject should remain consistent in time and vary between subjects in a systematic way. We note a major drawback of the previous studies, as they tend to operate under that assumption, without validating it first. To alleviate this fundamental issue, we pose and answer the following research hypothesis: “How accurately the subject-specific OPMM can generalize to the novel data outside of its train set?”

Deadline: Dec. 3, 2021, midnight