KOREAN

Optim. Lab.

Jong-June Jeon jj.jeon@uos.ac.kr

Univsersity of Seoul, Deaprtment of Statistics

Welcome!! Welcome to Optim.Lab at the Department of Statistics, University of Seoul. Our research laboratory focuses on developing machine learning models, including various applications of neural networks, and studying optimization methodologies for fitting these models. Currently, our lab mainly deals with image, network, weather, ranking, and stock market data. We work on tasks such as feature vector extraction, probability structure estimation, and the development of classification and regression models within these data domains. You can explore the research papers produced by our lab through the following Google Scholar page.

Recent Work (2023-Present)

Cho, S., Hong, S., and Jeon, J. J. (2024), Adaptive Adversarial Augmentation for Molecular Property Prediction (submitted)
An, S., Hong, S., Jeon J. J. (2024), "Distributional Learning for Synthetic Tabular Data via Cramer-Wold Device", (submitted)
Hong, S., Jeon J. J. (2024), "Uniform Pessimistic Risk and Optimal Portfolio", (submitted) (Link)
Hong S., An, S. and Jeon J. J. (2024), "Improving SMOTE via Fusing Conditional VAE for Noise Adaptive Sampling", (submitted)
Hong, S., Choi Y. , Jeon J. J. (2023), "Interpretable Transformer for Water Level Forecasting" , (under revision) (Link)
An, S., Song.K,, Jeon J. J. (2023), Causally Disentangled Generative Variational AutoEncoder", ECAI 2023 (Link)
An, S. and Jeon J. J. (2023), "Distributional Variational AutoEncoder To Infinite Quantiles and Beyond Gaussianity", NeuRIPS 2023 (Link)
Oh C., So J., Byun H., Lim Y., Shin M., Jeon J., Song.K (2023), Geodesic Multi-Modal Mixup for Robust Fine-Tuning, NeuRIPS 2023
An, S. and Jeon (2023), "Customization of Latent Space in Semi-Supervised Variational Autoencoder" , Pattern Recognition Letters (Link)
Kim D., Cho. S. Jeon J. J. and Choi J. (2023), "Screening inhalation toxicity of environmental chemicals using OECD test guideline 403, 412 and 413 data-based machine learning classification models", (submitted)

On-Going Work

Hong S., An, S. and Jeon J. J. (2023), Multivariate Cryptocurrency Price Forecasting using Distributional Variational AutoEncoder
Park, J. and Hong S., Cho, Y., and Jeon, J. J. (2024), A framework of Dynamic Neural Networks for Sea Ice Prediction
Choi, S., Park, T. and Jeon, J. J. (2024), Deep learning for predicting Traffic noise
An, Shin, and Jeon, Sustainable ML-pipeline for customizing LLM
Moon, H., An, S., and Jeon, J.J. (2024), Stopping Exploration improves bandit algorithm?
Lim, J., Woo, K., An, S., and Jeon, J.J. (2024), End-to-End Sythetic Data Generation under missing data
Cho, S., Shin, W., Woo, K., Kim, K., and Jeon, J.J. (2024), Link prediction model for CTD data analysis

Funding

Personalized Research on Large Language Models, 2023, University of Seoul
Development of Robo-Advisor Algorithms, 2023, Gwanak Research Institute
Basic Research Lab, 2022-2024, Ministry of Science and ICT
DS+ Project Team, 2022-2025, Ministry of Science and ICT
Digital Innovation Talent Development Project (Big Data), 2021-2023, Ministry of Education
Basic Researcher Support Project, 2022-2024, Ministry of Science and ICT
Development of Single Household Estimation Models, 2021-2023, Seoul-UOS-SKT-KCB-Shinhan Card Consortium
Research on Molecular Toxicity Networks, 2021-2025, Ministry of Environment
Development of Artificial Intelligence Decision Models, 2021

Contact

Building: Room 712, Mirae Hall, Universitu of Seoul
tel: 82-2-6490-2637 / fax: 82-2-6490-2629
Address: Department of Statistics, University of Seoul, 163 Jeonnong-ro, Dongdaemun-gu, Seoul, 02504, South Korea

Social Network

LAB.Current Research

There is a brief description of ongoing research with lab members. The research topic for 2023 is Causal Inference.

Deep Learning Molecular Structure

Members: S. Hong (PhD), S. Cho, H. Choi

Methodology: Causal Inference, GNN, Seq2Seq, Attention mechanism, Transformer, Bert

Applications: Recommendation System, Keyword extraction, Document summarization, Dynamic programing

Funding: Ministry of Environment

Leader: S. Cho

Causally Entangle Generative Model

Members: S. Hong (PhD), S. An, S. Choi

Methodology: Causal Inference, Variational Auto Encoder, Scoring methods (normalizing flow), Bayesian Statistics

Applications: Image generation, object detection, image segmentation

Funding: Ministry of Science and ICT

Leader: S. An

Ranking algorithm for information retrieval

Members: J. Jung, W. Shin

Methodology: Ranking Model

Applications: Information retrieval

Funding: -

Leader: JaeHwan Jung

Urban Big Data

Members: H. Moon, J. Park

Methodology: Causal Inference, Spatial data analysis

Applications: population, business model, policy etc.

Funding: Korea Credit Bureau

Leader: Huiwon Moon

Lab. Curriculum

Here's a list of fundamental computer, mathematics, and statistics topics that students in the Master's and Ph.D. programs at Optim Lab typically need to learn. Through regular student and professor seminars, you will learn various concepts essential to data science and acquire skills for their practical application.

Graduate school entrant (Requirements)

Python/R Basic (Undergraduate)
Probability (Undergraduate)
Linear Algebra (Undergraduate)
Regression Analysis (Undergraduate)
Mathematical Statistics (Undergraduate)

※ 통계학과 대학원 특별 세미나 (winter school) 계획 (매년 2월 15-19일, 6시간×5일간)

Graduate Student (M.S. and Ph.D.)

Multivariate Statistics (PCA, Copula)
Machine Learning (likelihood model, empirical risk minimization, Multivariate Calculus, Monte-Carlo Simulation)
Deep Learning (Python advanced, Database (SQL and transaction), pytorch (Custom layer, Custom loss, Gradient Tape))
Convex optimization (graduate course: optimization)

Graduate Student (Ph.D.)

Real Analysis (undergraudate)
Probability and general measure theory (graduate)
Asymptotics (graduate)
Empirical risk process

Oct, 2023

Education & Carrier

Director, Urban Bigdata & AI Institute (2023-)

Vice director, Office of research affairs, Univsersity of Seoul (2021-2022)

General Director, Convergence and Open Sharing System (Bigdata), Univsersity of Seoul (2021)

General Director for Statistics and Data, Seoul Metropolitan Government (2019-2020)

PHD

Doctor of Philosophy in Statistics (Thesis: High dimensional rankding model)

Department of Statistics, Seoul National University

Aug, 2012

BS

Bachelor of Business Administration

School of Business Administration, Seoul National University

Aug, 2005

LAB.Q&A

(Q) What kind of topics do students engage in when they come to the research lab?
(A) At Lab. Optim., students primarily understand linear algebra and optimization methodologies. They learn computational methods to fit machine learning models or artificial intelligence models. After developing the ability to interpret machine learning models as representations of mathematical models, they read papers in their field of interest and implement methodologies. For master's students, this might involve collaborating with companies on projects, developing lecture materials, or assisting in teaching. Ph. D. students focus on cultivating more profound mathematical abilities and working on the lab's priority projects. Refer to the Research menu for the lab's priority projects in 2023.
(Q) Are you recruiting undergraduate interns?
(A) We have internship programs within the school's curriculum or at the research lab level for undergraduate students. Please contact us via email.
(Q) What types of projects can students work on and how can they participate?
(A) While the projects vary each year, they mainly revolve around machine learning and artificial intelligence applications. In 2023, projects include maritime vessel control, chemical toxicity prediction using molecular structures, and analysis of single-person households using fusion data. To participate in projects, foundational research skills such as data preprocessing, analysis, and report writing are necessary. Participation is decided through consultation between professors and researchers at the project's outset, and tasks are assigned based on individual capabilities.
(Q) What kind of work are graduates mainly engaged in after graduation?
(A) Most graduates are involved in data analysis-related roles. They are actively continuing their engagement in various fields such as finance, distribution, and life sciences, performing data analysis tasks.
(Q) Is there financial support available for master's and doctoral programs?
(A) Scholarships within University of Seoul, scholarships from the Talent Development Project, and financial support for lab projects can be provided for both master's and doctoral programs. Please refer to the Funding section for more information.
(Q) Are there non-regular programs available as well?
(A) Please inquire through the Department of Statistics.
(Q) What are the requirements for a Ph.D. graduation?
(A) To graduate with a Ph.D., you need to be the first author of either one SCI(E) paper or two KCI papers.

Lab.Family

Post Doc.

Sungchul Hong: Optimization in Group lasso, and Quantile regression, Reinforcement learning (Home)

Ph. D. Students

Soyoung Cho: Bayesian Network in medical field, Deep learning for toxicology
SeungHwan An: Image generation (Home)
Wonhyung Shin: Neural Ranking and Collaborative Filtering
JaeHwan Jung: Data fusion, Kalman filter
강민설: (현: 언어모형 DB)
이수재: (현: 경기도청)
장홍진: (현: 신용보증기금)
김영욱: (현: 품질관리)

M. S. Students

Huiwon Moon: Collaborative Fitering with bandit algorithm
Haein Choi: Machine Learning for toxicity analysis
Sung-Su Choi: Knowledge Distillation
김기훈: -
정지훈: -
조윤서: -

Undergraduate Research Internship

박새빈: Seoul citizen life sytle data analysis (2023 fall)
이경윤: Transformer Forecaster for finance (2023 fall, mentor: 최해인)
신현서: Long term forecasting with deep learning models (2023 spring, mentor: Kyungju Gil )
박기정: Long term forecasting with deep learning models (2022 winter, mentor: Soyoung Cho)
박재성: Long term forecasting with deep learning models (2022 winter, mentor: Soyoung Cho )

졸업생

Sangjun Moon (Ph.D.): Hana Card ( Personal Home)
Sungchul Hong (Ph.D.): Post Doc.
Sungjun Byun (M.S.): DREAMCIS
Heegun Kang (M.S.): Ph. D. student, UOS
Sanghee Chae (M.S.): TBA
Eunyoung Ko (M.S.): LOTTE CARD
Yi Jeong Han (석사): GS RETAIL
Yeonsun Cho (M.S.): KDATA
Jiwon Lee (M.S.): SAERON SOLUTION
Kyuchul Eum (M.S.): SK C&C
Duhee Han (M.S.): LG CNS (Data Platform task)
Yeju Hong (M.S.): PT-KOREA
Hyukgu Kim (M.S.): KB KOOKMIN BANK
Jina Park (M.S.): DATA SOLUTION
Kyungju Gil: Generative model for zero inflated data

Useful Links

LAB.News

Institution of Continuing Education : 2023. 8
- 4주차
- 5주차
Parallel GPU Computing : 21 Aug. 2023
- Concept and Practice
Reinforcement learning : Apr-June. 2023
- Prof. Jisu Kim (UNIST)
- Prof. Jungseul OK (POSTECH)
- Prof. Min-hwan Oh (SNU)
Causal Inference : 4차 (22시간)
- 인과추론소개: Purdue University 정용한
- 준모수추론이론: 건국대학교 유규상 교수님
- Causal inference with score matching: 서울대학교 이권상 교수님
- 공정한 AI: 서울대학교 김용대 교수님

Summer, 2022

Basic Research : Robust VAE Model Study using Distributionally Robust Learning (2022-2024)(2022-2024)
- This research aims for theoretical analysis of Distributionally Robust Optimization (DRO)
- Researchers: Master's and Doctoral Students

May, 2022

Basic Research Lab (Graduate School): Trustworthy AI Model Research using Causal Inference (2022-2024) - news!!
- A research team composed of Professors Choi Yeon-jin from the Department of Statistics, Song Kyung-woo from the Department of Artificial Intelligence, and Jung Ji-young is conducting research on trustworthy AI models using causal inference.
- Participating Researchers: 1 Postdoctoral Researcher, 3 Doctoral Students, 17 Master's Students, 1 Bachelor's Student

May, 2022

Data Science Convergence Project (Graduate School): DS Plus Project (2022-2028) - news!!

May, 2022

Weekly Seminar Presentation by Research Lab: Regular seminar presentations by the research lab. The seminar results from a total of 15 sessions will be presented on topics related to Vision, Text, and Urban Big Data at a conference room in Gwanghwamun on February 17, 2022. We hope that 2022 will mark the transition from being a Research Follower to a Leader group.

Jan, 2022

Regularization of Weekly Seminar in the Research Lab: Small research groups within the lab hold regular seminars. These seminars cover topics such as reading recent statistical papers, implementing object detection models, researching keyword and sentence extraction algorithms, and conducting research related to urban big data.

Oct, 2023

Digital Innovation Shared University Talent Development Project - Big Data Selection (2021-2026): Digital Innovative Shared University for the Talent Development of New Digital Technologies (Big Data) - news!!
- A consortium of Seoul National University, Gyeonggi Science and Technology University, Gyeongsang National University, University of Seoul, Sookmyung Women's University, Chonbuk National University, and Handong University has been selected for the undergraduate Big Data-related curriculum.<

May, 2021

자료분석학회 2023 summer, SeungHwan An (박사과정) , 학생논문상 우수상 축하합니다!
Distributional Learning of Variational AutoEncoder: Application to Synthetic Data Generation.
자료분석학회 2023 summer, Soyoung Cho (Ph. D. student) 학생 논문 장려상 축하합니다!
Robust Graph Augmentation for Molecular Property Prediction.
자료분석학회 2023 summer, 길경주 (석사과정) 학생 포스터 논문상 우수상 축하합니다!
Zero-inflated Data Generation using TVAE.

July, 2023

SeungHwan An (박사과정) ECAI 2023 accept을 축하합니다!

July, 2023

Sungchul Hong(박사) 졸업, 김혁구(석사) 졸업, 박진아(석사), 한두희(석사) 졸업을 축하합니다!

Feb, 2023

Sungchul Hong 학생의 한국인공지능학회(2022 Fall) 우수논문상 수상을 축하합니다! (Uniform Pessimistic Risk and its Applications, Hong and Jeon)

Nov, 2022

김혁구, 박진아 학생의 서울데이터펠로우십 선정을 죽하합니다!- news!!

May, 2022

고은영, 한이정, 조연선 졸업을 축하합니다

June, 2019

한국 통계학회에서 발표 (2019년 추계): 전종준, 문상준, 홍성철, 한이정이 발표하였습니다 (click!) .

Nov, 9, 2019

수상소식!!: 2019년 추계 한국통계학회에서 문상준 (학생발표 1등), 홍성철 (학생발표 2등), 한이정(학생발표 3등) 축하합니다!!!

Nov, 15, 2019