2:00pm - 2:30pm

Dr. Damian Valles Molina

Electrical & Computing Engineering, Ingram School of Engineering, Texas State University

Title: Development of Emotion Recognition with ML\ DL Techniques and New Challenges

Abstract: The unmet service tool for children with autism spectrum disorder (ASD) is the lack of tools that help them recognize or teach them to understand human emotions. Like many children with ASD, the target population prefers or focuses on handheld devices that provide graphical interphases that capture their attention. Our current effort is developing an app tool that will help them recognize and understand the human emotions of people with whom they interact daily. Emotions are classified and detected by facial, speech, and body-gesture motions when interacting with someone. The app will be supported via Machine and Deep Learning models that discretely recognize emotions based on different human traits. The app will respond and provide an emoticon to the screen to indicate the recognized feelings based on the AI models’ overall outcomes. Over time, the app’s utilization can offer children and adolescents a tool to develop skills and identify patterns in the emotional state of people they interact with using models that account for different ethnic backgrounds. The recognition and understanding over the life span can provide better communication with parents and caretakers at a level that might have been otherwise difficult to convey if someone was happy or upset with them or about anything else. The overall objective is to develop a tool available for all to use on mobile or handheld devices and improve human emotions’ communication and recognition skills over time. However, these models provide a biased and usually incorrect recognition of how people interpret emotions. This is because datasets are typically represented with actors or exaggerated samples. Emotions also depend on cultural backgrounds and other aspects of our brain that help interpret complex behaviors. Our study will continue to explore the environment, timing, and other features that help recognize human emotions.