Aarhus University Seal / Aarhus Universitets segl

Special talk by Beril Sirmacek on Trusting on AI for better health

2019.09.30 | Søs Küster Markussen

Date Fri 11 Oct
Time 10:00 11:00
Location Ada-333


My passion is to develop fully automated computer vision and AI frameworks which can support prediction and early detection of disease or other patterns which are interesting to be detected for decision making. In this talk, I will introduce how AI frameworks can support medical doctors for helping larger numbers of patients in shorter time with higher confidence. In some fields, especially in the medical field, failure or dysfunction of the computer vision algorithms is not an option. Thus, trusting on AI and explaining the decision making of the ‘black-box’ structure becomes an important issue. Therefore, I would like to introduce some early disease prediction and decision making examples, and discuss interpretability measures for developing better AI frameworks for better health services.

Short Bio: I received my PhD degree in 2009, officially from the faculty of Electrical Engineering and Computer Science in Istanbul. During my PhD, I have worked as a teaching assistant at University of Trento and Technical University of Munich. After my PhD degree, I continued living in Munich. I have worked at German Aerospace Center (DLR) Earth Observation Group. During my years in Germany, I worked as a guest lecturer at several German universities. In 2012, I started to work at TUDelft. After the end of my contract, I focused on developing my own AI and big data processing framework (farmAR app) which has been awarded by EU Commission twice, for highly supporting the sustainable development goals. Since 2017, I have been with University of Twente, at Robotics and Mechatronics group, where I develop AI, computer vision and SLAM algorithms in the field of healthcare. Currently, I am also teaching the Optimal estimation for dynamic systems course at MSc level. My personal website: www.BerilSirmacek.com

CS frontpage, Public/media, Target groups