Myeonggu Kang
Master Student, Industrial Artificial Intelligence Laboratory
Kyung Hee University
Research Interests: AI-Driven Intelligent Manufacturing & Process Intelligence.
Focus: ML/DL, Time-Series Forecasting, GNN & RL, LLM (RAG/FT).
Education
M.S. in Artificial Intelligence
2025.09 - Present
Kyung Hee University
- Research Assistant, Industrial Artificial Intelligence (IAI) Lab (2024.01 - 2025.08)
B.S. in Industrial and Management Systems Engineering (Double Major)
Graduating 2025.08
Kyung Hee University
B.S. in Software Convergence
Graduating 2025.08
Kyung Hee University
- Academic Excellence Award (Top Student), College of Engineering (2024 - Senior Year)
- Academic Excellence Award (Top Student), College of Engineering (2023 - Sophomore Year)
Projects
Human-Robot Collaboration Boundaries: Analysis of Preferences for Intelligent Robots
2025.05 - 2025.11
Best Paper Presentation Award | 2025 KSOI Autumn Conference
- Conducted a study on user acceptance of intelligent robots in Industry 5.0 using Conjoint Analysis and Discrete Choice Models (DCM).
- Analyzed 500 respondents' preferences regarding work allocation, safety systems, and connectivity.
- Identified significant preferences for "Human-Supervised" collaboration models and real-time monitoring capabilities.
Manufacturing-Specialized SLM for Integrated Equipment & Process Monitoring
2025.03 - Present
- Collaborative R&D project with an industrial automation company.
- Developing a domain-specific Small Language Model (SLM) system for intelligent factory monitoring.
Korean Sarcasm Detection Model & Corpus Construction
2025.03 - 2025.07
Team Leader | Poster at HCLT-KACL 2025
- Built a Korean sarcasm classification system combining rule-based filtering and LLMs.
- Collected and refined datasets from social media and online communities.
- Used LangChain for prompt engineering to generate structured sarcastic dialogues.
- Fine-tuned and evaluated KoBERT and BLOSSOM-3B models.
Multi-step Wind Power Generation Prediction Using TimesFM
2024.09 - 2025.06
Independent Researcher | Poster at KCC 2025
- Leveraged Google's TimesFM (Foundation Model) for renewable energy forecasting.
- Integrated 4 years of wind turbine data and weather forecasts from Jeju Island.
- Implemented multi-step forecasting (24 hours ahead) using historical data and exogenous features.
Maritime Object Detection on Ships Using YOLO
2024.09 - 2024.12
- Conducted object detection research using maritime images captured on ships.
- Contributed to model tuning and detection performance improvement using YOLOv8.
Sleep Stage Classification & Image Generation with Multimodal Biosignals
2024.03 - 2024.09
Team Leader | Poster at KSC 2024 (Encouragement Prize)
- Developed a CNN+LSTM model for sleep stage classification using EEG, EOG, EMG, and ECG signals.
- Implemented full-stack pipeline: Signal preprocessing (Wavelet denoising) to Model training.
- Created a "Dream Visualization" system using Stable Diffusion (SDXL-Turbo) based on REM-stage features.
Wind Power Forecasting System Using Time-series DB
2024.03 - 2024.09
Presented at Korea Smart Media Association Autumn Conference 2024
- Designed an end-to-end architecture using Machbase-NEO (TSDB) for efficient data handling.
- Implemented LSTM models with walk-forward validation.
- Demonstrated significant reduction in system overhead compared to offline processing.
Object Detection on 2D Road Scene Images
2023.09 - 2023.12
Team Leader | Collaboration with Gyeonggi Autonomous Driving Center
- Trained YOLOv5m-based model for pedestrians, vehicles, and cyclists detection.
- Managed labeling analysis, preprocessing, and hyperparameter tuning.
Selected Publications & Presentations
An LLM-based Question Answering System over Relational Databases via Text-to-SQL
Oral Presentation at 2025 SEBS Fall Conference (The Korea Society for Electronic Commerce) | 2025.11.29 - Best Paper Award
View Proceeding
Human-Robot Collaboration Boundaries: Analysis of Preferences for Intelligent Robots
Oral Presentation at 2025 KSOI Autumn Conference | 2025.11.21 - Best Paper Presentation Award
View Proceeding
A Data Corpus Generation Pipeline for Advancing Korean Sarcasm Detection
Poster Session at HCLT-KACL 2025 (Daejeon, Korea) | 2025.10.1~2
Multi-step Wind Power Generation Prediction Using a Time-series Foundation AI Model
Poster Session at KCC 2025 (Korea Computer Congress, Jeju Island) | 2025.07/1~4
Deep Learning-Based Sleep Stage Classification Using Multimodal Biosignals
Poster at KSC 2024 (Korea Software Congress, Yeosu) - Encouragement Prize | 2024.12.18~20
Development of Machine Learning-based Wind Power Forecasting Systems Using Time-series DB
Oral Presentation at 2024 Autumn Conf. of Korea Smart Media Association (Seoul) | 2024.10.11~12
Activities & Global Experience
Research Assistant, IAI Lab
2023.12 - 2025.08
- GNN Study: Covered GCN, GAT applications in manufacturing fault diagnosis.
- LLM & RAG: Explored Retrieval-Augmented Generation pipelines and fine-tuning.
- LangChain: Study group on LLM orchestration and tool integration.
- XAI: Studied SHAP, LIME, and model-agnostic interpretability.
University of Auckland (New Zealand) - ELA Program
2025.02
Kyung Hee University Major Study Abroad Program
- Completed intensive English training and Project/Risk Management studies.
Kyung Hee University Data Analytics Club (KHUDA)
2024.02 - 2024.12
- Awards: Best Study Group Prize (ADsP certification group).
- RL Track: Led hands-on projects including Unity ML-Agents simulations.
- Events: Organized intercollegiate Datathon; Attended Gyeonggi Great Global Transformation Forum.
Certifications & Skills
Certifications: ADsP (Advanced Data Analytics Semi-Professional) - KData (2024)
Core Expertise: AI (Machine Learning, Deep Learning, LLM)
Research Focus: Time-Series Forecasting, GNN, Reinforcement Learning, RAG/Fine-Tuning
Core Expertise: AI (Machine Learning, Deep Learning, LLM)
Research Focus: Time-Series Forecasting, GNN, Reinforcement Learning, RAG/Fine-Tuning