Research

Publications

• Topic : Spectral Bias in Physics-Informed Neural Networks

  Youngsik Hwang, Kangjun Lee, Jeongsik Yoon, Dongyoung Lim† Under review · 2026

Projects

SVM

2025.11 – 2025.12

SKiP: SVM weighted by K-Nearest-Neighbors and class Probability for weakening outliers

Machine Learning Course Project · UNIST

Implemented a KNN and probability-weighted SVM to handle feature and label outliers. Collaborated as a four-person team.

PINN

2025.07 – 2025.08

PINN for Lithium-ion Battery SOH Prediction

Machine Learning & Optimization Lab · UNIST

Applied PINN architecture for predicting state-of-health of lithium-ion batteries using the NASA battery dataset with added ICA feature.

RAG

2025.04 – 2025.06

RAG Improvement via LoRA Fine-Tuning

NLP Course Project · UNIST

Trained a small GPT model and RAG pipeline, then applied LoRA-based parameter-efficient fine-tuning using the Self-RAG framework.

SAM

2025.01 – 2025.02

Sharpness-Aware Minimization: Implementation and Evaluation

Machine Learning & Optimization Lab · UNIST

Implemented SAM in PyTorch, trained Wide-ResNet and custom CNN on CIFAR-10. Extended to MultiSAM with multi-step perturbations and compared with SGD/SAM baselines.

HCI

2024.03 – 2024.06

Smart Attendance System

HCI Course Project · UNIST

Invented and developed an attendance checking system from a human-computer interaction perspective. Collaborated as a four-person team.