RESEARCH INTERESTS & ACTIVITIES

▶ 주요 연구 주제 (TBU)

∙     Univariate/Multivariate Change Point Detection

∙     Concept Drift Detection Techniques

∙     Meta Learning for Segmentation


∙     Novelty Detection via Autoencoder Variants

∙     Clustered Multi-task Sequence-to-Sequence Learning

∙     Model-Based Transfer Learning

∙     Meta Learning (Deep Seed Region Growing Segmentation)


∙     Network-Based Causal Analysis and Interpretation

∙     Explainable Artificial Intelligence (XAI) for Unstructured Datasets (Image, Video)

∙     Collaborative Learning Techniques without Centralized Training Data


∙     Self-Adaptive Learning on Non-stationary Environments

∙     AI-Based Dynamic Dispatching: Production Scheduling Problem, Dispatching Rule Selection

∙     Machine Learning-Assisted Optimization: Shortest Path Search, Dynamic Vehicle Routing, Facility Location Problem


∙     Nurse Workforce Planning and Strategy

∙     AI-Based Medical Institution Evaluation

∙     Recommendation System for Healthcare Optimization 

∙     Invasive Ductal Carcinoma Detection via Image Segmentation Techniques

∙     Development of Upper Body Exercise Equipment via Human Pose Estimation (HPE)



∙     Structure Learning of Bayesian Networks

∙     Graph-Regularized Deep Neural Networks

∙     단변량 / 다변량 변화탐지

∙     Concept Drift 탐지 기법

∙     이미지 세그먼테이션 위한 메타학습 기법


∙     오토인코더 기반의 재구축 방식 이상탐지

∙     군집화 기반 다중 시퀀스-투-시퀀스 학습

∙     모델 전이학습

∙     DSRG 기반 메타학습


∙     네트워크 모델 기반 인과분석 및 해석

∙     비정형 데이터 대상 설명 가능한 인공지능 (XAI)

∙     협업필터링 기반 학습법


∙     비정상성 환경을 고려한 자율 학습 기법 연구

∙     인공지능 기반의 디스패칭 최적화 연구

∙     기계학습 기반의 최적화 문제 해결 연구: 최단거리 탐색, 동적 차량 경로탐색, 설비위치 결정 문제 등


∙     간호학 응용 분야

∙     최적화 및 인공지능 기반 의료기관 평가문제

∙     의료기관 추천시스템

∙     병리 이미지 분석 기반의 전이성 유방암세포 탐지

∙     인간자세추정(HPE) 모델에 기반한 상지운동기기 개발


∙     베이지안 네트워크의 구조학습

∙     그래프 기반 딥러닝 모델의 정규화 방법론

▶ 연구 과제 소개 (TBU)

eXplainable AI for Evaluating Tourist Satisfactions for Korean Local Festivals

This study predicts the traveler satisfaction for local festivals and and identifies significant issues for festival program. We use the predictive and surrogate models to obtain both highly accurate performance and understandable results.

Lightweight, eXplainable, and Self-Adaptive Learning for Real-time Multivariate Analysis in Smart Manufacturing

To leverage the applicability of AI to the real-world plants, we propose promoting three significant techniques such as 'Lightweight AI', 'eXplainable AI', and 'Self-Adaptive AI.' We expect that these studies will provide a major turning point to give critical opportunities for the effective use of AI in intelligent manufacturing.

Moral concerns about data integrity for AI's Ethical Applications

In South Korea, Lee Luda, an AI chatbot, has been pulled after inappropriate dialogues such as abusive and discriminatory expressions and privacy violations. We need to discuss the AI's ethical applications and thus secure the well-being of our society with the rapid advancement of AI. (with “SECURE Team For You” (SweEt spot ConsUlting REsearch Team For the next generation, You) 

Computer Vision Techniques in Cancer Pathology 

We here proposes various machine learning methods to predict and highlight automatically the malignant tumors in histology image. Regarding histology image analysis, we propose using a semantic segmentation technique with the state-of-the-art approaches in self-supervised learning, semi-supervised learning, and anomaly detection tasks.

Transfer Learning for Clustered Multi-Task Sequence-to-Sequence Learning

This study resolves a low resource machine learning problem to obtain predictive control of idle vehicle repositioning to maximize machine utilization 

Road Abnormality Detection by Using Weakly-Supervised Semantic Segmentation

The proposed approach detects the abnormal objects including pothole in road by using the seeded region growing module to benefit from deep features with small samples

Parallelized Metaheuristics for Stochastic Probabilistic Models

This study focuses on parallelizing the focuses on parallelizing-based graph search for structure learning of Bayesian networks

AI-Based Adaptive Control for Autonomous Vehicle Systems in Semiconductor Plants

AI-Based adaptive control system is proposed to improve the effectiveness of the Autonomous Vehicle Systems.  Async-HTTP server and client pair is implemented to achieve a non-blocking prediction messaging service

Concept Drift-Based Incremental Learning for Robust Autonomous Vehicle Control

The proposed method combines a drift-adaptation learning technique with a drift detector to achieve adaptive traffic prediction in time-varying Autonomous vehicle systems

Simulation-Optimization with Hybrid Metaheuristics for Robust Decision Making

This study proposes a hybrid approach combining genetic algorithm, depth first search, and dynamic programming to approximate the optimal positions for the location of facilities in plants. Further, discrete event simulation is used for evaluation.