ECE Seminar: Yasin Abbasi-Yadkori

Starts:
11:00 am on Wednesday, February 21, 2018
Ends:
12:00 pm on Wednesday, February 21, 2018
Location:
Photonics Center, 8 Saint Mary's Street, Room 339
URL:
http://www.bu.edu/eng/files/2018/02/YasinFlyer.jpg
ECE Seminar: Yasin Abbasi-Yadkori Wednesday, February 21, 2018 Yasin Abbasi-Yadkori Data Scientist Adobe Photonics Center, 8 Saint Mary's Street, Room 339 Refreshments will be available at 10:45am outside of PHO 339 Title: Learning and Planning in Sequential Decision Problems Abstract: Many decision problems have an interactive nature; the decision maker executes an action, receives feedback from the environment, and finally uses the feedback to improve the next decision. Such sequential decision problems are particularly challenging when the decision and state spaces are large, which is often the case in many areas such as robotics, healthcare, and finance among others. In this talk, I will present my research in planning and learning in large sequential decision problems. The studied problems range from basic linear regression problems, to more complicated problems with limited feedback such as bandit linear optimization and linear quadratic (LQ) control, to problems that are both computationally and statistically challenging such as reinforcement learning and Markov decision processes. I will present algorithms that are provably data-efficient and can be executed in real-time. The proposed algorithms rely on efficient exploration of the state and action spaces. I will discuss an approach to the design of efficient exploration methods. The method that I demonstrate involves construction of tight confidence sets for linear regression, and design of fast mixing Markov chains. Bio: Yasin Abbasi-Yadkori is a Data Scientist in Adobe Research, San Jose. Prior to that, he was a postdoctoral researcher in Queensland University of Technology with Peter Bartlett. He received his Ph.D. in Computing Sciences from University of Alberta in 2012 under the supervision of Csaba Szepesvari. He is broadly interested in developing autonomous and adaptive agents that perform well in challenging environments.