Estimating Ego States: The Machine Learning Perspective

Abstract

The interface between Artificial Intelligence (AI), machine learning (ML) and psychology is an intensively explored research area. Specifically, Transactional Analysis (TA), with its structured and precise language, presents a promising area for applying ML techniques, unveiling new potential research avenues. This article explores the intersection of artificial intelligence, machine learning, and psychology, focusing on developing a method and the software environment for estimating ego states using MS Kinect™ sensor data. The research investigates the application of TA theory, emphasizing capturing the behavioural ego indicators. While Kinect skeletal data is considered, the gestures and postures are the primary input. The Authors present an innovative approach to annotate and visualize Kinect data using video streams for further autonomous ego state estimation. Within this study, they collected a dataset of 15 students from The Silesian University of Technology. The data was acquired through the use of both a video camera and a Kinect sensor. The nine distinct labels were employed for data annotation. They reflect Parent, Adult, and Child ego states across different temporal dimensions encompassing the past, present, and future. The study includes preliminary results demonstrating the outcomes of this approach's visualization technique and their interpretation. The final part of the article discusses the potential of applying the presented method in applications for the education field.