Tutorial Lecture 1: Saturday, 2 November, 9 am – 10:30 am @ Chung Kunmo Conference Hall, 5th floor, bldg #E9, KAIST

Youngchul Sung, KAIST, Korea

Biography: Youngchul Sung received B.S. and M.S. degrees in Electronics Engineering from Seoul National University, Seoul, Korea in 1993 and 1995, respectively. After working at LG Information and Communications, Ltd. as a senior research engineer from 1995 to 2000, he joined the Ph.D. program and received Ph.D. degree in Electrical and Computer Engineering with Applied Mathematics as minor from Cornell University, Ithaca, NY, USA in 2005. From 2005 until 2007, he was a senior engineer in the Corporate R & D Center of Qualcomm, Inc., San Diego, CA, USA and participated in design of Qualcomm’s 3GPP R6 WCDMA base station modem. Since 2007 he has been on the faculty of the School of Electrical Engineering in KAIST, Daejeon, South Korea. For more biography, click here.

Exponential Family, Divergence and Related Geometry
Abstract: Kullback-Leibler divergence is the universal divergence widely used in statistics and machine learning. It is always non-negative and is zero when two distributions are the same. Hence, it is considered as some kind of “distance” between two points in the space of probability distributions. This tutorial aims to provide a basic understanding of Kullback-Leibler divergence. The tutorial starts with maximum entropy principle and resulting exponential family of probability distributions. Then, it explains the dual geometry generated from the cumulant generating function of exponential family and the associated Bregman divergence, which coincides with Kullback-Leibler divergence in the case of exponential family, and explains the generalize Pythagorean theorem in the dual geometry. Furthermore, we deal with the invariance property of divergence.

Tutorial Lecture 2: Saturday, 2 November, 2 pm – 3:30 pm @ Chung Kunmo Conference Hall, 5th floor, bldg #E9, KAIST

Jaegul Choo, Korea University, Korea

Biography: Jaegul Choo is currently an assistant professor in the Dept. of Computer Science and Engineering at Korea University. He received M.S in the School of Electrical and Computer Engineering at Georgia Tech in 2009 and Ph.D in the School of Computational Science and Engineering at Georgia Tech in 2013, advised by Prof. Haesun Park. From 2011 to 2014, he has been a research scientist at Georgia Tech. During the summer in 2009 and 2010, he worked at National Visualization and Analytics Center (NVAC) in Pacific Northwest National Laboratory. He earned his B.S in the Dept. of Electrical and Computer Engineering at Seoul National University. For more biography, click here.

Image-to-image translation via generative adversarial networks
Abstract: This talk will introduce state-of-the-art image-to-image translation approaches based on generative adversarial networks, including pix2pix, CycleGAN, and StarGAN. I will then conclude the talk with other advanced models as well as future research directions.