Hello! I'm Jin Choi!
I am a Ph.D. student in the College of Computing
at Georgia Tech advised by professor Haesun Park.
My research interests include NLP and Visualization,
specifically in Interactive Learning and Incremental
Learning.
My goal is to bridge humans and AI to improve
machine performance by utilizing human expertise.
Experience
- AI Research Engineer Intern
@ Upstage AI, Korea, Summer 2021
- Research Scientist Intern
@ Megagon Labs, Mountain View, CA, May 2020 - Aug 2020
- Mentors: Çağatay Demiralp, Sara Evensen
- Proposed a system for data programming by demonstration for span-level annotation
- Graduate Research Assistant
@ Georgia Institute of Technology, Atlanta, GA, Aug 2018- Present
- Undergraduate Research Assistant
@ Seoul National University, Seoul, Korea, Aug 2016 - Jan 2018
Educations
@ Georgia Institute of Technology, Atlanta, GA, USA, Aug 2018 - Present
- B.S. in ECE(Minor: CSE)
@ Seoul National University, Seoul, Korea, Mar 2011 - Feb 2018
Publications
HITL-NLP Survey
Putting Humans in the Natural Language Processing Loop: A Survey
Zijie J. Wang, Dongjin Choi, Shenyu Xu, Diyi Yang
EACL Workshop on Bridging HCI and NLP, 2021
[PDF]
S3CMTF
S3CMTF: Fast, accurate, and scalable method for incomplete coupled matrix-tensor factorization
Dongjin Choi, Jun-gi Jang, U Kang
First parallel and scalable method for sparse coupled matrix-tensor factorization (Sparse CMTF) with HOSVD format.
PLOS ONE, 2019
[PDF] [Github]
SNeCT
SNeCT: Scalable Network Constrained Tucker Decomposition for Multi-Platform Data Profiling
Dongjin Choi, Lee Sael
Network constrained Tucker decomposition on multi-platform genomic data.
IEEE TCBB, 2019
[PDF] [Github]
Zoom-SVD
Zoom-SVD: Fast and Memory Efficient Method for Extracting Key Patterns in an Arbitrary Time Range
Jun-gi Jang, Dongjin Choi, Jinhong Jung,
U Kang
Proposed a framework to store singular value decomposition (SVD) of multiple sensor data block-by-block and efficiently retrieve the SVD of an arbitrary time range query.
CIKM, 2018
[PDF] [Homepage]
CTD
CTD: Fast, Accurate, and Interpretable Method for Static and Dynamic Tensor Decompositions
Jungwoo Lee, Dongjin Choi, Lee Sael
First interpretable sampling-based dynamic tensor decomposition method. We demonstrate the usefulness of CTD by detecting distributed denial of service (DDoS) attack.
Received bronze medal from Samsung humantech paper award.
PLOS ONE, 2018
[PDF] [Github]
Activity Prediction
Activity Prediction from Sensor Data using Convolutional Neural Networks and an Efficient Compression Method
Woojeong Jin, Dongjin Choi, Youngjin Kim, and U Kang
Isolated kernel CNN model for people flow recognition.
KIISE, 2018
[PDF]
Projects
OCR Annotation Tool
Internship Project in Upstage AI,
May 2021 - Aug 2021
- Boosted OCR Annotation process with the interactive user interface by validating and annotating labels at the same time
DebateVis
Information Visualization term project (CS7450), Aug 2019 - Dec 2019
- Interactively visualized 2019 democratic presidential debate
- Try Demo
People Flow Recognition
People flow recognition for intelligent building systems (IBS), Sovico, Samsung, Sep 2017 - Jan 2018
- Implement a pedestrian simulator (watch simulation video)
- Proposed isolated kernel CNN model for people flow recognition
- Proposed multi-scale skip connected and graph-structured RNN model for people flow prediction
Deep Learning Term Project
Deep learning term project, Seoul Nathional University, Mar 2017 - June 2017
- Term project for graduate deep learning course @SNU
- Developed CNN model for human identification on image using transfer learning technique
Awards & Honors
- Humantech Paper Award (Bronze prize, Co-author)
@ Samsung, Feb 2017
- Included in the top 6 in 'Computer Science & Engineering' division
- Best term paper award in 'Writing in Science & Technology'
@ Seoul National University, Dec 2015
- paper (written in Korean)
- 1st place among 50 teams
- The paper proposed a new design for motorcycle helmet shield to reduce driver's air pollutant inhalation
- I drew the blueprint for 3-D printing and implemented Arduino experiment code
Patents
- Data Analysis Method and Apparatus for Sparse Data
- U Kang, Dongjin Choi, Jun-Gi Jang
- Korean patent number: 10-2017-0158496, (filed on Nov, 24, 2017).
Miscellaneous
- Mentoring program for Dream Camp
- Seongwon High School, Namwon, Korea, Summer 2015
- I served as an advisor and mentor for students of Seongwon high school
- Volunteer work as one of the co-founders of Teach for Korea
- Seoul, Korea, 2011-2012
- I was a full-time tutor for high school students in a low-income family
- I served more than 300 hours of total
- English debate TA for GLPS
- Hoengseong, Korea, Summer 2011
- My high school hosts a summer vacation camp, GLPS, for elite young students
- I worked as a TA for English debate
- Volunteer work as the leader of For & From Them
- Rania Mission School, Kolkata, India, Summer 2011
- I taught various activities (origami, mathematical quizzes) to young students in the low-income family
- I served as the manager for the volunteer work, I managed overall plans and schedules for the work
- Jochiwon, Korea, Winter 2010
- I attended the winter school for algorithms and combinatorics as an only high school student
© Jin Choi 2021
Computer Science Researcher
jin.choi@gatech.edu
jinchoi.xyz
Last updated: 1/19/2022