Sangmin Lee

Machine Learning Engineer

"저는 Machine Learning Engineer을 목표로 경기대학교에서 공학사, 이학석사를 취득하였고, 해당 과정을 통해 연구 경력은 4년을 쌓았으며 다양한 분야의 역량 향상을 위해 노력해왔습니다.
경험을 통해 쌓은 문제해결역량을 통해 산업 발전에 기여하는 것이 목표입니다."

Technical Writer @Slashpage

Education

I am pleased to introduce my educational background.

Kyonggi University / M.S. of Computer Science (2022.03 ~ 2024.02)

G. P. A: 4.25 / 4.5

Transcript: https://drive.google.com/file/d/1M9FXnkug1mlgOMzFjaqaePkw-dTLapX-
Graduate Certificate: https://drive.google.com/file/d/1MAzl9yFkpjk7_8OnHp7P3RMP70OTpuuW
Research Field : Network, Deep Learning (Recommendation System, Computer Vision, Graph Neural Network)
Advisor : Prof. Namgi Kim (Kyonggi university)

Kyonggi University / B.S. of Computer Engineering (2015.03 ~ 2022.02)

G. P. A: 3.62 / 4.5

Transcript: https://drive.google.com/file/d/1M787Fnq8NTA9b-NVVdzjAuyuUKxQutPD
Graduate Certificate: https://drive.google.com/file/d/1M98UlQGH_CCHMAJs8Lhc2LCZxLdBqfc7
Research Field : Deep Learning(Computer Vision)

Honors & Awards

The honors I have had and the awards I have received are as follows.

Top 0.04%

DACON Competition Rank

11th / 104,243th

(2024.03.26)

Top 4.41%

DACON Code Sharing Rank

110th / 2,495th

(2024.02.27)

212 EA

Programmers Ranking

18,378th, 1,421 point

(2024.01.27)

87,863th

BAEKJOON Rank: Silver 5

(2024.01.27)

Participation

Kyonggi University Department of AI and Computer Science

No. 2021-0005

Best Paper

Paper: Health Trainer Web Services with PostNet

KIIT, No. 21-222

Publication

The results of my efforts to build artificial intelligence capabilities are as follows.

Master’s thesis

Optimization methods of Graph Convolution Networks for Recommendation Systems

Master’s thesis, UCI: I804:41002-000000057783, 22 February 2024.
dCollection: http://www.dcollection.net/handler/kyonggi/000000057783, 1st author.

International journal

Deep Learning Model Ensemble for the Accuracy of Classification Degenerative Arthritis

SCIE(Q2)-Computers Materials & Continua (CMC), vol. 75, no. 1, pp. 1981-1994, 2023.
DOI: https://doi.org/10.32604/cmc.2023.035245, 1st author.

Weighted Forwarding in Graph Convolution Networks for Recommendation Information System

SCIE(Q2)-Computers Materials & Continua (CMC), vol. 78, no.2, pp. 1897–1914, 2024.
DOI: https://doi.org/10.32604/cmc.2023.046346, 1st author.

Embedding Enhancement method for LightGCN in Recommendation Information Systems

SCIE(Q3)-KSII Transactions on Internet and Information Systems(TIIS), Reviewing.
DOI: , 1st author

Learning optimization method for Graph Neural Network-Based Graph Convolutional Network Optimization

SCIE(Q2)-Institute of Electrical and Electronics Engineers, Incorporated Access, Editing.
DOI: , 1st author

International conference

Egress Initialization for Graph Convolution Network in Recommendation Systems

•

The 15th International Conference on Internet (ICONI 2023), 1st author.

•

Paper: https://drive.google.com/file/d/1NOZhATzb0IiM_4R9Q1kPt3Q9iUbmxgLW

Thin Graph Convolution Network in Recommendation Systems

•

The 15th International Conference on Internet (ICONI 2023), 1st author.

•

Paper: https://drive.google.com/file/d/1NOuPfeNwhXyLCmEkHJJZrL7TvkTqdA7K

Over-smoothing in LightGCN doesn’t happen much

•

The 15th International Conference on Internet (ICONI 2023), 1st author.

•

Paper: https://drive.google.com/file/d/1NY5f-mC0upgpc6RrEzRSM0LLlvHM8pT-

Leverage time-based data to improve recommendation system accuracy

•

The 18th Asia Pacific International Conference on Information Science and Technology(APIC-IST), 1st author

•

Paper: https://drive.google.com/file/d/1NpnzKgpIXi9hj3ppNdHPUZ_Z1uul6kXE

Using Deep Learning for Medical Automation in Diagnosis of Degenerative Arthritis

•

The 8th international Conference for Small and Medium Business 2023(ICSMB), 1st author

•

Paper: https://drive.google.com/file/d/1NkRMAtIfoBin7RqoAcx3uSSTXPkHTLTq

Classification of degenerative arthritis using Xception model in radiographic images

•

The 17th Asia Pacific International Conference on Information Science and Technology(APIC-IST), 1st author

•

Paper: https://drive.google.com/file/d/1NnIDjIx2_sI47TagEIhxfezrRhfNMO8f

•

This paper was selected as one of the best papers of APIC-IST 2022 and recommended for publication in CMC-Computers, Materials & Continua (CMC).

Domestic journal

Analyzing the use of time-based user behavior data to improve the performance of graph neural network-based recommendation systems

•

KCI-The Journal of Korean Institute of Information Technology(JKIIT), vol. 21, No. 5, pp. 21-28, 2023, ISSN: 1598-8619 (Print) 2093-7571 (Online)

•

DOI : https://doi.org/10.14801/jkiit.2023.21.5.21, 1st author

Improved Pancreas Segmentation using Multiple Concatenated U-Net Model for Medical Image Systems

•

KCI-The Journal of Korean Institute of Information Technology(JKIIT), vol. 20, No. 5, pp. 81-87, 2022, ISSN: 1598-8619 (Print), 2093-7571 (Online)https://doi.org/10.14801/jkiit.2022.20.5.81, 1st author

•

DOI: https://doi.org/10.14801/jkiit.2022.20.5.81, 1st author

Domestic conference

Analysis of performance of unsupervised anomaly detection methods according to anomaly score threshold selection criteria

•

KSII Summer Conference, 2023, 2st author

•

Paper: https://drive.google.com/file/d/1NIr8MxxH8JAoxbwJsOt9i0YaEzMe2tkv

Analysis of Traffic Flow Prediction Performance According to Spatial-Temporal Attention Module Structures in GNN-based Deep Learning Model

•

KSII SPRING Conference, 2023, 2st author

•

Paper: https://drive.google.com/file/d/1NGGQBCDCT_oFME2FsqLbGo4FdPtrk2qz

Time-based data Analysis for improving recommendation system performance

•

KSII SPRING Conference, 2023, 1st author

•

Paper: https://drive.google.com/file/d/1NER2aVumQcQB_KhpF22YXf7wNW7mjgCp

A Study on the Flow Classification Method Using Auto-Encoder Model in SDN Environment

•

KICS Winter Conference, vol.80, 2023, ISSN 2383-8302, 2023, 3st author

•

Paper: https://drive.google.com/file/d/1N8e6NUy7Qo80dZykoEiFSgVkHW02kuI4

Optimizing caching servers using a recommender system

•

SEBS Fall Conference, 2022, 1st author

•

Paper: https://drive.google.com/file/d/1N4MvBWUqUenyc99KjX_l64IOtZxyEcup

Implementation of attendance system through RFID tags

•

KSII SPRING Conference, 2022, 3st author

•

Paper: https://drive.google.com/file/d/1MviIHIG6TfhDdKguF70f5P4XSoqd8F6w

Accuracy analysis of convolutional neural network models for overlapped character image classification

•

KSII Fall Conference, 2021, 1st author

•

Paper: https://drive.google.com/file/d/1MvdD6_Md7dS1X9yA-cH_MxtGLLM2sOAv

Health Trainer Web Services with PostNet

•

Proceedings of KIIT Conference, pp. 729 - 733. 2021.06, 1st author

•

Paper: https://drive.google.com/file/d/1MvL7frWl2c1aJuPDspkuF-J4_2EkhG3

•

KIIT Best Paper Award

Project

The projects to develop my software skills are as follows.

💬

Optimizing GCN for Recsys (23.01 ~ 23.12)

WF Technique, Egress Technique, Non-combination architecture

•

Role: Project leads and researchers

•

Research Field : Graph Convolutional Network, Recommendation Systems, Optimization techniques.

•

Project result: https://github.com/d9249/WFGR, https://github.com/d9249/OptGCN

•

Description: During the second year of his master's program, he proposed WF technique, Egress technique and Non-combination architecture for optimizing graph convolutional networks in recommendation systems, and improved learning speed and accuracy through this project.

🗃️

Smart I.o.T Lab Homepage (21.12 ~ 22.04)

•

Role: Front Developer

•

Skill : Vue.js, Node.js, JavaScript

•

Project Result: https://netlab.kyonggi.ac.kr/

•

Description: Creation of the Smart I.o.T Lab website and addition of features commissioned by Prof. Nam-Ki Kim

🧞‍♂️

IA^2S (Image Analysis Automation System)
(21.10 ~ 22.02)

•

Role: Team Leader, Front Developer

•

Skill : JavaScript, Django, OpenCV, Python 3.7.

•

Project Result: https://github.com/HHFEHH/crack-automation

•

Description: Web services for crack flattening, crack length measurement, and data automation to improve existing methods for physical exterior wall repair and building maintenance companies.

📔

Fit-sibang (21.03 ~ 21.06)

•

Role: Team Leader, Front Developer

•

Skill : React.js, Node.js, JavaScript

•

Project Result: https://github.com/KGU-Code-15/fit-sibangI

•

Description: Made as a web app, it can be accessed on any device and helps you exercise through posture estimation using AI.

External Activities

Activities to develop my communication and research skills are as follows.

💬

LG Aimers (24.01 ~ 24.02)

•

Role: 4th semester.

•

Link: https://github.com/d9249/LG_Aimers

•

Research Field: Deep Learning

•

Description: LG Aimers is LG Group's youth education program that provides AI education and the opportunity to experience participating in AI hackathons that deal with LG's actual data.

🗃️

Pseudo-lab (22.09 ~ 22.11)

•

Team: 그래프로 설득하기.

•

Link:https://www.notion.so/chanrankim/ac12010e21af4431a5f3e8ae2bdc4ee3?pvs=4

•

Research Field : Graph Neural Network

•

Description: Conduct projects in their respective domains that utilize Graph to solve real-world needs, from defining the problem to why the need should be met with Graph.

🧞‍♂️

DIYA (21.03 ~ 23.02)

•

4th Team: Computer Vision

•

Role: Team members

•

Research Field : Deep Learning, Computer Vision(paper review, participate in competition).

•

5th Team: Computer Vision, Graph Neural Network, Pytorch

•

Role: Team members

•

Research Field : Computer Vision, Graph Neural Network, Deep Learning.

•

Certificate: https://drive.google.com/file/d/1NxSxsWUAXOWbk2WgKtwxHrU4dsBsbY8Y

💡

CS224W-KOR (22.07 ~ 22.09)

•

Role: Team members

•

Research Field : Graph Neural Network.

•

Project Result: https://cs224w-kor.github.io/blog/

•

Description: CS224W, presenting and organizing the parts they studied and translating them into Korean.

📔

7th college student coding camp
(21.06 ~ 21.07)

•

Certificate: https://drive.google.com/file/d/1NzwgCVIDXgRZPjK3mbp-L94mf2fR9vRY

Competition

The list of artificial intelligence competitions I participated in to improve my problem-solving skills is as follows, and I have organized the competitions that ranked in the top 10%.