University of Alicante
Artificial Intelligence Applied to Teaching and Learning Processes
Abstract
This doctoral dissertation addresses the personalization of the educational process in in-person teaching environments through the analysis of visual and biometric data, exploring the application and evaluation of Artificial Intelligence (AI) systems in face-to-face settings. Despite the great potential of AI in in-person education, its adoption is limited by factors such as algorithmic bias, privacy risks, and high implementation costs. To explore practical solutions, this thesis develops a low-cost architecture for the capture and monitoring of students in face-to-face classes. Subsequently, the resulting monitoring software is validated using the Technology Acceptance Model (TAM) among early-career engineering faculty. As a primary contribution, this architecture was employed to create two new datasets, DIPSEER and CADDI, which were specifically designed to train and validate future AI models in this area with visual and biometric information from students in in-person classes. Building on this foundation, the IPSEER dataset was validated using Large Language Models (LLMs) models, including their visual (VLMs) and multimodal (MLLs) variants, as advanced tools for measuring attention and emotion. Additionally, the potential of these models to simulate educational role-playing behaviors and generate realistic interactions was investigated, thereby highlighting the need to include new visual and in-person-focused datasets for a comprehensive behavioral simulation of such roles. Finally, the appendix presents the collaborations undertaken with various researchers during the course of this doctoral dissertation
Author and committee
dc:creator, dc:contributor.*- Author
-
- Márquez Carpintero, Luis
Subjects
dc:subject × 7Identifiers
dc:identifier.*- Identifier
- hdl:10045/167870
- OAI identifier oai:identifier
- oai:rua.ua.es:10045/167870