Title: Measuring autonomic arousal in exposure therapy: wearables, virtual reality, and uncertainty
Abstract:
As deep learning becomes more ubiquitous in modern day applications, particularly in the field of health and medicine, the issue of overconfident predictions associated with frequentist strategies in neural networks (NN’s) needs to be addressed. In this work we demonstrate steps towards applying Bayesian approaches in deep NN's to understanding and quantifying autonomic arousal in human patients undergoing virtual reality exposure therapy.
