B.S. Candidate, Computer Engineering Gachon University · Seongnam-Si, South Korea
I am Gyuyeon Lim, a research intern currently studying time-series AI and NLP. In addition to my research, I have been working as a lead mathematics instructor for Korean CSAT (College Scholastic Ability Test) preparation since 2023.
Get in touch
Education


Research
Learning compact, structured embeddings from high-dimensional data. Focuses on contrastive and self-supervised methods, data embedding, and latent space geometry.
Temporal and spatio-temporal modeling for forecasting and classification of multivariate sensor streams. Covers sequence models, graph-based dependencies, and large-scale time-series architectures.
Applying deep learning to physical and biological signals — IMU, EEG, and multivariate sensor data. Integrates physics priors, domain constraints, and signal structure into model design.
Publications
Rotation-Robust Self-Supervised Learning for IMU-Based HAR with Multi-Scale Convolutional Backbones
Robust Sensor-Based Activity Recognition via Physics-Aware Disentangled Representation Learning and Feature-wise Linear Modulation Conditioning
CAP-SSM: Context-Aware Prototypes and Selective State-Space Modeling for Efficient Wearable Human Activity Recognition
Prototype-Guided Physics-Modulated Perceiver for Human Activity Recognition
Contrastive Learning for Zero-Shot Human Activity Recognition Using Labeled Simple Actions on Wearable Devices
Experience




Activities
Gachon University — The Highest Honor
Gachon University
Ministry of Science and ICT (MSIT)
Anyang-Gwacheon Chamber of Commerce and Industry
Career Exploration Credit Program · Gachon University · GitHub
2025 Yonsei Yongin Severance Hospital Healthcare Hackathon
Gachon University · YOLOv5 Children Protection Area detection · GitHub
Team GPT — 168th / 376 teams · KINTEX, Goyang-Si · Official
Gachon University (1st Place)
Gachon University (1st Place) · Smart Home with Physical Computing
2018 National Student Science and Invention Competition
Projects
Zero-shot HAR with contrastive learning on wearable devices.
github.com/lky473736/ZSL-HARPrototype-Guided Physics-Modulated Perceiver for HAR.
github.com/lky473736/P2-PerceiverContext-Aware Prototypes + Selective State-Space Model for wearable HAR.
github.com/lky473736/CAP-SSMRotation-robust SSL via multi-scale convolutional backbones.
github.com/lky473736/InceptionMKActivity-Based Feature Augmentation with Multi-Kernel TCN + Transformer.
github.com/lky473736/ABFA-HARPhysics-Aware Disentangled Representation with FiLM conditioning.
github.com/lky473736/PADRAnomaly detection using Diffusion, DDIM, DDPM on fall detection datasets.
github.com/lky473736/diffusion_DDPM_fall_detectionPersonal AI study archive — 101 to expert level (2024–now).
github.com/lky473736/learning-aiAll records of paper work and research readings.
Notion →Real-time one-click ZOOM access program for Hyosung High School.
github.com/lky473736/scarecrow