Attention performance classification based on eye tracking and machine learning

Visual attention represents a key concept in cognitive performance and visual information processing. The relationship that exists between cognition, problem-solving, and attention can be explained and modeled through the use of eye-tracking technologies. Novel technologies like machine learning can enhance cognitive performances by finding gaze patterns and using them to direct and guide attention in problem-solving scenarios. Our experiment aimed to classify attention into three performance classes: low, average, and high, using eye tracking data. For the task at hand, the most relevant eye-tracking features found were mean gaze velocity, saccade and fixation count, and average fixation duration. Our study showed that machine learning models trained on eye-tracking features can classify attention performance with over 81% accuracy.