NSF East Asia and Pacific Summer Institute (EAPSI) for FY 2013 in Japan

Steinmetz Erik S, Minneapolis MN

Investigators

Abstract

This action funds Erik Stefan Steinmetz of The University of Minnesota, Department of Computer Science and Engineering, to conduct a research project in the Computer and Information Science and Engineering area during the summer of 2013 at The Nara Institute of Science and Technology (NIST) in Nara, Japan. The project title is "Improving Opening Strategy in Computer Go Players." The host scientist is Kenichi Matsumoto. This project explores a method of improving the artificial intelligence search technique known as Monte Carlo search. The avenues being explored in this research allow the application of expert knowledge to a technique which normally relies on not being restricted by contextual knowledge (to allow randomness to properly explore a search tree). Preliminary results showed one such measure which improved play of a computer Go engine that used Monte Carlo search, and this project develops further measures along with discovering their effectiveness and limitations. By studying these modifications which touch the very nature of Monte Carlo search, this project explores the applicability and limitations of Monte Carlo search in general and whether or not hybrid methods offer an opportunity for their improvement. This research will be directly applicable to other problems using the Monte Carlo search method such as military simulation software, along with some large online simulation software systems. Broader impacts of an EAPSI fellowship include providing the Fellow a first-hand research experience outside the U.S.; an introduction to the science, science policy, and scientific infrastructure of the respective location; and an orientation to the society, culture and language. These activities meet the NSF goal to educate for international collaborations early in the career of its scientists, engineers, and educators, thus ensuring a globally aware U.S. scientific workforce. Furthermore, the demonstration of the results of this project will be exceptionally helpful in a classroom setting. It shows a simple method to increase the effectiveness of an algorithm in a topic which is interesting to undergraduate students. By showing how key ideas such as this research can be applied directly to these interesting topics, students can be engaged and motivated at a very early stage of their undergraduate computer science careers.