발표자: 서창호 (KAIST 교수) 발표일: 2017.5. Changho Suh is an Ewon Associate Professor in the School of Electrical Engineering at Korea Advanced Institute of Science and Technology (KAIST). He received the B.S. and M.S. degrees in Electrical Engineering from KAIST in 2000 and 2002 respectively, and the Ph.D. degree in Electrical Engineering and Computer Sciences from UC-Berkeley in 2011, under the supervision of Prof. David Tse. From 2011 to 2012, he was a postdoctoral associate at the Research Laboratory of Electronics in MIT. From 2002 to 2006, he had been with the Telecommunication R&D Center, Samsung Electronics. Dr. Suh received the 2015 IEIE Hadong Young Engineer Award, a 2015 Bell Labs Prize finalist, the 2013 IEEE Communications Society Stephen O. Rice Prize, the 2011 David J. Sakrison Memorial Prize (the best dissertation award in the UC-Berkeley EECS Department), and the 2009 IEEE ISIT Best Student Paper Award. 개요: Recommendation systems come up in a wide variety of applications like Netflix, YouTube, e-commerce, search engine, AI-assistance, to name a few. There has been a proliferation of recommender algorithms. One such prominent algorithm is based on matrix completion of which the goal is to reconstruct missing entries of a matrix from partially observed entries. In this talk, I will demonstrate that the matrix completion algorithm can play a role also in the context of education, serving as an AI-tutor that educates individual students by recommending a proper sequence of problems tailored for the understanding level of each student. Specifically I will introduce a matrix whose (i,j)-entry represents the probability of student i giving a correct answer to question j, and show that the solution of matrix completion can help predicting answers for unsolved questions and thus can be utilized for problem-recommendation. We conduct an extensive set of experiments based on a large educational data set (consisting of millions of TOEIC test responses from tens of thousands of students) to demonstrate that our algorithm can reliably predict the understanding level of individual students, outperforming a conventional logistic regression approach aided by experts. Our technique has been implemented in an online education platform collaborating with an edu-tech startup company named Riiid. I will also discuss our recent on-going approach based on deep learning. Our preliminary results show that the new approach is superior than the earlier one, having the potential to give impacts upon other numerous applications.