This was a part of final project in Artificial Intelligence class.
Given the task of choosing two datasets from UC-Irvine Machine Learning Repository, I used "Student Performance" and "Turkieye Student Evaluation" as the two data sets. The specific requirements for the project were as follows:
[Dataset] # of Attributes (10 ~ 100)
[Dataset] # of Instances (100 ~ 1,000, Greater than 1,000)
[Analysis] Use at least 4 different classifiers for comparative analysis
[Analysis] The final evaluation must use cross-validation
[Analysis] Make sure to use the concept of overfitting in final evalution
[Analysis] Use the results of zeroR, oneR as the basepne
In this work, I analyzed business districts in Seoul using sales data.
I used python for pre-processing and analysis. All the work was done in Jupyter Notebook.
From November of 2010 to January of 2013, I have taught more than 2,000 students in person (cf. here). Although I started teaching without much expectation, I realized that I love teaching and creating curriculum for students. Also, the experience has imbued me with insight for learning experience (LX). Ever since, I have consciously and unconsciously conducted various researches by experimenting with various tools and methods for a great LX. Some of the resulting products include MoviEng, ALGOGA, Pronunciation Clinic and Speaking without Thinking.
As a student researcher at Data and Visual Analytics (DAVIAN) Lab at Korea University, I worked on building a new algorithm based on Nonnegative Matrix Factorization (NMF) under Prof. Choo 's supervision. The proposed method discovers high-quality local topics, which solves the problem inherent in approaches geared towards discovering global topics that are often rather meaningless and redundant.
The paper has been accepted to ICDM'16! (Best Student Paper Award, 8.5% acceptance rate) [paper]
As a student researcher at Data and Visual Analytics (DAVIAN) Lab at Korea University, I worked with a group of students under Prof. Choo 's supervision to create a visualization system for Convolutional Neural Network (CNN). The paper has been accepted to Interactive Data Exploration and Analysis (IDEA) workshop at KDD'16 and its extended version to Future of Interactive Learning Machines (FILM) workshop at NIPS'16. The following is the poster presented at the IDEA workshop. [paper]
EYEscort is a navigation application for Android that assists the blind when crossing the road. It utilizes beacon technology to locate the position of the blind while crossing and if the user begins to deviate from the crosswalk, it gives warning through bluetooth headset or vibration. The work was advised by Prof. Kim and was awarded with a prize from SK Creative Challenge Program held at HCI Korea 2016. The following is a poster which was presented at HCI Korea 2016. Other HCI-related works I have done are as follows: , , , , 
Cognitive Science plays a crucial role in my research interests, as it impacts the way people learn, experience and perceive. The following essay is a part of the work done for "Intro. to Brain Science" class and discusses whether glial cells are merely helper cells for neurons.
Previously a former student at the University of Chciago with an aspiration for studying economics, I am a keen observer of human nature and society, and I believe that conducting research with consideration of the evolving dynamics between people and society is important. The following essay is a part of the research done for "A World without Discrimination" class. Recognized as "one of best papers" among the research papers submitted by more than 100 students in the class, I was asked by a professor to present my research in front of the class. The research is on 'Abuse and Discrimination of Non-regular Employment,' written in Korean. [ppt]