Ph.D./M.Sc. Development of Machine Learning Algorithms for In-Ear Biosignal Monitoring

Application deadline: Wed 01 Jun 2022

The goal of this project is to provide usable and robust biosignal captures in the occluded earcanal for various applications, ranging from health monitoring, hand-free/silent interfaces to cognitive biometry (resulting from emotion recognition capabilities). Currently, heartbeat and breathing rates were successfully extracted from in-ear recordings [1]. A corpus of in-ear recorded Wearer Induced Disturbances (WID) was also used to train a classifier that was successfully used to detect a wide range of non-verbal audio events (clicking of teeth, clearing the throat, saliva noise, coughing, talking, etc.) [2]. The student will work on improving the real-time robust classification of in-ear biosignal using machine learning approaches, merging HRV measurement and non-verbal event classification, thereby enabling the multimodal characterization of emotions, cognitive performance, and cardiovascular health from in-ear biosignals and non-verbal events.

More Information in this PDF file.