Jonas Nordhaug Myhre - Postdoctoral researcher with the Machine Learning group at UiT Data Science
Date: - Location: Eurecom
Title: Controlling blood glucose levels in T1 diabetic patients with fitted Q iterations and Gaussian Processes Abstract: In this talk I will present ongoing work in Reinforcement Learning and Type 1 Diabetes. Type 1 Diabetes is characterized by the lack of insulin-producing beta cells in the pancreas. The so-called artificial pancreas promises to alleviate the burdens of self-management. While the physical components of the system -- the continuous glucose monitor and insulin pump -- have experienced rapid advances, a technological bottleneck remains in the control algorithm. This work presents a fitted Q reinforcement learning algorithm based on functional features and Gaussian Process regression. Biography: Jonas N. Myhre is a postdoctoral researcher with the Machine Learning group at UiT - The Arctic University of Norway. He is currently working on developing and testing deep reinforcement learning methods for controlling blood glucose levels in people that suffer from Type 1 Diabetes. In 2009 he received the Bsc in Automation Engineering, in 2011 the Msc in Electrical Engineering, and in 2016 the PhD in machine learning, all from UiT The Arctic University of Norway. His research interests include manifold learning, deep reinforcement learning, kernel methods, random projections and manifold optimization.