Research and Work Experience

Data Mining LaboratorySeoul National University, Seoul, Korea

Research Intern (Advisor: Prof. U Kang) $\cdot$ Jan. 2016 - Present

• Research area: network representation learning, probabilistic modeling, and deep learning

WISEfn CorporationSeoul, Korea

Software Engineer $\cdot$ July 2012 - May 2015

• Developed investment strategy evaluation software and workflow management system
• Managed database management tools and data integration procedures

Nonlinear Neurophysics LaboratoryKAIST, Daejeon, Korea

Research Intern (Advisor: Prof. Soo Yong Kim, Kyungsik Kim) $\cdot$ Mar. 2007 - Feb. 2008

• Research area: random matrix theory, network theory, and game theory

Publications

SIDE: Representation Learning in Signed Directed Networks

We propose SIDE, a network embedding algorithm for signed directed networks. Network embedding learns a mapping of each node to a vector. SIDE carefully formulates and optimizes likelihood over both direct and indirect signed connections. We provide socio-psychological interpretation for each component of likelihood function and prove linear scalability of our algorithm.
In WWW (To Appear), 2018.

UniWalk: Explainable and Accurate Recommendation for Rating and Network Data

We propose UniWalk, an explainable and accurate recommender system that exploits both social network and rating data. UniWalk combines both data into a unified graph, learns latent features of users and items, and recommends items to each user through the features. It provides the reason why items are recommended together with the recommendation results.
In arXiv:1710.07134, 2017.

Structure of a financial cross-correlation matrix under attack

We investigate the structure of a perturbed stock market in terms of correlation matrices. Through statistical analyses such as random matrix theory (RMT), network theory, and the correlation coefficient distributions, we show that the global structure of a stock market is vulnerable to perturbation. However, apart from in the analysis of inverse participation ratios (IPRs), the vulnerability becomes dull under a small-scale perturbation. When going down to the structure of business sectors, we confirm that correlation-based business sectors are regrouped in terms of IPRs.
In Physica A, 2009.

Patents

. Explainable and Accurate Recommender Method and System using Social Network Information and Rating Information. In KR: 10-2017-0159167, 2017.

. Apparatus and Method for Representation Learning in Signed Directed Networks. In KR: 10-2017-0130914, 2017.

Projects

Transfer Learning Algorithm Design

Designing a novel transfer learning algorithm for deep neural network

Efficient Graph Mining using CUR Decomposition

Designed an algorithm that decompose graph adjacency matrix by leveraging graph structure

Signed Graph Generative Model

Designing novel transfer learning algorithm for deep neural network

Social Recommender System

Designed a novel recommendation model leveraging social network by network embedding

Deep Learning for Stock Price Prediction

Developed stock price prediction software using deep neural network

Deep Learning Library Development using TensorFlow

Developed a library implementing high-level neural network API

Awards and Honors

• Bronze Medal, National Collegiate Mathematics Competition (for math major), Korea Mathematical Society, 2015
• Silver Medal, National Collegiate Mathematics Competition (for non-math major), Korea Mathematical Society, 2011
• National Scholarship for Science and Engineering, Korea Student Aid Foundation, 2009-2015
• Silver Medal, Korea Physics Olympiad, Korea Physical Society, 2008
• Merit-based Scholarship, Kwangshin Petroleum Co., Kyobo Life Insurance Co., Chokwang Paint Ltd., 2006-2008

Teaching

• Teaching Assistant of Basic Calculus 2 at Seoul National University, Fall 2010
• Teaching Assistant of Basic Physics 1 at Seoul National University, Spring 2010

Invited Talks

Machine Learning and Deep Learning with Stock Price Data

Apr. 22, 2016, Invited talk, Emoney Inc.