07 Aug Multi-modal behavioral modeling in assistive technologies: Enhancing human performance through sensorial sensing
Thu, September 3, 2020 6PM
Recent technological advances have led to an unprecedentedly easy access to cost-effective but high-end sensors that came to stay as part of our daily lives. From smart-phones and watches to off-the-shelf brain-wave monitoring devices and robotic platforms, ubiquitous and pervasive computing was never more realistic than it is today. This evolution has allowed the emergence of several interdisciplinary studies that blend together diverse domains like computer science and AI with psychology, psychiatry and other medical fields. The purpose of most such studies could be summarized to the following three points: a) provide a better understanding of common human conditions, b) automate detection and predict future events and c) provide valuable feedback to the users in order to improve performance and well-being. The targeted conditions can have a detrimental toll to our psyche and body and usually include stress, depression, fatigue, attention deficit and others and are linked to various disorders and diseases like ADHD or Multiple Sclerosis. This talk aims to present a diverse set of studies and applications of how multi-modal systems can enhance human performance under the umbrella of the aforementioned topics. From data-collection and data cleaning, to analysis and modeling we discuss the challenges and the potentials of machine learning for discovering specific behavioral characteristics and learning personalized and interpretable results.
ABOUT THE SPEAKER
Michalis Papakostas is a Postdoctoral Researcher working with the Language and Information Technology group at University of Michigan’s AI Lab. He received his Ph.D. in 2019 from the University of Texas at Arlington and his interests and expertise are focused on multimodal processing for human behavior modeling and monitoring with applications in a variety of areas mainly related to assistive technologies.