Learning & Decision
- Reinforcement learning for sequential decision-making
- Self-play and multi-agent interaction
- Actor-critic methods and policy optimization
Sangmyung Artificial Intelligence Laboratory
SMAI Lab
- Principal Investigator
- Prof. Min-Suk Kim
- Affiliation
- Sangmyung University
- Student Lab
- Hakmugwan M405
- Reinforcement learning and decision-making
- Deep learning for vision and time-series data
- Applied AI systems in simulation and field data
SMAI Lab
Artificial intelligence methods for applied systems.
SMAI Lab studies reinforcement learning, deep learning, generative models, and representation learning, and applies these methods to simulation, vision and time-series data, industrial systems, and robotic control.
Methods
RL · DL · Generative AI
Data
Vision · Time Series · Simulation
Systems
Robotics · Industry · Security
M405
Student Lab, Hakmugwan
Research
Research Areas
SMAI Lab develops AI methods for decision-making, perception, representation, and applied intelligent systems. The lab works across algorithms, data-driven modeling, and real-world applications.
Anomaly & Perception
- Representation learning from vision and sensor data
- Time-series modeling for system-state understanding
- Multimodal perception for complex observations
Generative & Representation
- Generative modeling for data-limited settings
- Prompt automation and model adaptation
- Knowledge transfer and reusable representations
Applied AI Systems
- AI methods for simulation and diagnostics
- Learning-based control and intelligent systems
- Applied AI for industrial, robotic, and security contexts
Projects
Research Projects
Active Projects
Core Technologies for Self-Evolving AI-Based Cyber Attack-Defense Systems
Ministry of Science and ICT · Mar. 2022 - Dec. 2027
Remote Diagnostics for Secondary-System Equipment Using Ultra-Low-Power Wireless Monitoring
Ministry of Science and ICT · Apr. 2022 - Dec. 2029
Completed Projects
Multi-Agent Self-Play Reinforcement Learning for Cyber Attack Response Planning
Ministry of Science and ICT · Feb. 2022 - Feb. 2025
Integrated Multi-Spectral Video Analysis and Surveillance Device Development
Ministry of Science and ICT · Mar. 2021 - Dec. 2023
Publications
Publications & Awards
SCI / SCIE
Extended Maximum Actor-Critic Framework Based on Policy Gradient Reinforcement for System Optimization
Jung-Hyun Kim, Yong-Hoon Choi, You-Rak Choi, Jae-Hyeok Jeong, Min-Suk Kim
DOI 10.3390/app15041828Optimal Cyber Attack Strategy Using Reinforcement Learning Based on Common Vulnerability Scoring System
Bum-Sok Kim, Hye-Won Suk, Yong-Hoon Choi, Dae-Sung Moon, Min-Suk Kim
DOI 10.32604/cmes.2024.052375Leak Event Diagnosis for Power Plants: Generative Anomaly Detection Using Prototypical Networks
Jaehyeok Jeong, Doyeob Yeo, Seungseo Roh, Yujin Jo, Minsuk Kim
DOI 10.3390/s24154991Intelligent Complementary Multi-Modal Fusion for Anomaly Surveillance and Security System
Jae-hyeok Jeong, Hwan-hee Jung, Yong-hoon Choi, Seong-hee Park, Min-suk Kim
DOI 10.3390/s23229214KCI
생성형 모델 기반 이미지 변환 성능 향상을 위한 프롬프트 자동화 연구
최용훈, 김민석 · 28(2), 179-187
DOI 10.9717/kmms.2025.28.2.179Maximized-Actor-Critic(MAC)을 이용한 정책 경사 강화학습 방법
김정현, 최용훈, 김민석 · 27(3), 455-463
DOI 10.9717/kmms.2024.27.3.455Spatio-Temporal Graph Normalizing Flows (STG-NF) 기반 공공장소 이상 탐지 일반화에 관한 연구
최용훈, 정재혁, 김민석 · 27(2), 352-361
DOI 10.9717/kmms.2024.27.2.352네트워크 공격 시뮬레이터를 이용한 강화학습 기반 사이버 공격 예측 연구
김범석, 김정현, 김민석 · 22(3), 112-118
DOI 10.1003/JNL.JAKO202330357645069RGB 비디오 데이터를 이용한 Slowfast 모델 기반 이상 행동 인식 최적화
정재혁, 김민석 · 25(8), 1049-1058
DOI 10.9717/kmms.2022.25.8.1049Conference Proceedings
심층 강화학습을 활용한 비동기식 Self-Play 사이버 공방 연구
김정현, 정재혁, 장준원, 김민석 · 102-104
이미지 텍스처 특징에 기반한 센싱 데이터 이상탐지 연구
최용훈, 김민서, 김민석 · 우수논문상
강화학습을 활용한 MITRE ATT&CK 기반 네트워크 공격 시뮬레이션 개발
김범석, 김정현, 구기종, 김민석
Adaptive Reinforcement Learning Cyber Defense Strategies using CyberBattleSim
Bum-Sok Kim, Yonghoon Choi, Minsuk Kim · 150-151
Self-Play 사이버 공방을 위한 심층 강화학습 시뮬레이션 연구
김정현, 정재혁, 김민석 · 152-153
객체 이미지로 사전 학습된 PatchCore 기반 고위험 발전소 시계열 이상 탐지
최용훈, 조유진, 장준원, 김민석 · 270-271
사이버 공방 시뮬레이터를 이용한 강화학습 기반 에이전트 학습 개선 및 보상 재구성 연구
석혜원, 김민석 · 41-44
Mujoco 시뮬레이터를 활용한 로봇 제어 환경에서 강화학습 정책분석
장준원, 최용훈, 김민석 · 45-48
Cyber Attack and Defense Strategies using Reinforcement Learning via Simulated Network Environment
Bum-Sok Kim, Hye-Won Suk, Jung-Hyun Kim, Jae-Hyeok Jeong, Min-Suk Kim · NSR Best Paper
Enhancing Cyber Strategies through Reinforcement Learning: A CyberBattleSim (CBS)
Bumsok Kim, Minsuk Kim, Junghyun Kim, Yonghoon Choi, Seungseo Roh
Anomaly Detection using LSTM-Based Autoencoder with Time Series Dataset over a network
Hyewon Suk, Minsuk Kim
Study of Intelligent Cyber Range Simulation using Reinforcement Learning
Junghyun Kim, Bumsok Kim, Min-Suk Kim
Study of Technology for Anomaly Detection in Secure Edge System via Video Surveillance
Jae-Hyeok Jeong, Min-Suk Kim
Deep learning based Scheme for Developing Secure Systems in CCTV using Anomaly detection
Hyewon Suk, Junwon Jang, Minsuk Kim
Dynamic Monitoring and Surveillance of Anomaly Detection for Advanced Security Facility using Unsupervised Learning
Yonghoon Choi, Hwanhee Jung, Minsuk Kim · Sangmyung University
Awards
Best Paper Award
Undergraduate paper competition, 6th Korea Artificial Intelligence Conference
NSR Best Paper
MobiSec 2023 Best Paper · Cyber Attack and Defense Strategies using Reinforcement Learning via Simulated Network Environment
People
Members
Professor
Min-Suk Kim
Professor Lab · Hannuri Hall I508B AI methods and applied intelligent systems
Ph.D. Student
Jung-Hyun Kim
Ph.D. Student · Student Lab, Hakmugwan M405 Reinforcement learning and system optimization
M.S. Student
Yong-Hoon Choi
M.S. Student · Student Lab, Hakmugwan M405 Deep learning, vision, and time-series data
Undergraduate Researcher
Min-Seo Kim
Undergraduate Researcher · Student Lab, Hakmugwan M405 Cybersecurity simulation and applied AI
Undergraduate Researcher
Soo-Been Park
Undergraduate Researcher · Student Lab, Hakmugwan M405 Few-shot learning and computer visionNews
News
Sangmyung University media reported the Best Paper Award received by Yong-Hoon Choi and Minseo Kim at the 6th Korea Artificial Intelligence Conference undergraduate paper competition.
The undergraduate research work on image-texture-based sensing data analysis was introduced in external media as a Best Paper Award achievement.
SMAI Lab presented research on asynchronous self-play for cyber attack-defense simulation and sensing-data analysis at the 2025 Korea Artificial Intelligence Conference.
A reinforcement-learning-based system optimization study by SMAI Lab members was published in Applied Sciences.
SMAI Lab presented work on CyberBattleSim-based defense strategies, self-play cyber simulation, and power-plant time-series analysis at the 2024 Korea Artificial Intelligence Conference.
A generative anomaly-detection study for leak-event diagnosis in power plants was published in Sensors.
SMAI Lab presented research on reinforcement-learning agents for cyber simulation and MuJoCo-based policy analysis at the Korea Society of Computer Information summer conference.
SMAI Lab gave an oral presentation at MobiSec 2023 in Okinawa, Japan, and the paper received the NSR Best Paper Award.