Applied Statistics Major · Software Double Major · Chung-Ang University
I am interested in Foundation Models, Representation Learning, and Robust Adaptation under Distribution Shift.
My research experience spans ECG, EEG, PSG, audio, and speech, with a focus on how pretrained models adapt across changes in source, site, subject, session, and data conditions.
Research Interests
Foundation Models · Domain Generalization · Domain Adaptation ·
Parameter-Efficient Adaptation · Representation Steering · Cross-Domain Learning
PSG Foundation Model · Representation Steering · Domain Shift
- Built a downstream prediction pipeline using a pretrained PSG Foundation Model
- Applied SAE-based Representation Steering to investigate site-associated latent features
- Evaluated representation-level intervention under recording-site domain shift
EEG Foundation Model · Selective Unfreezing
- Applied a pretrained EEG Foundation Model for auditory attention decoding
- Implemented a 128 → 200 Hz TemporalResampler
- Compared 10% / 30% / 50% Selective Unfreezing
- Investigated audio-only → audio-visual condition shift
ECG · Cross-Source Generalization
- Developed a 1D ResNet-BiGRU model with physiologically motivated ECG features
- Designed and implemented RBBB/LAFB-related handcrafted features
- Investigated robustness under heterogeneous source distributions
- Extended the work to a journal paper as a co-first author
H. Im*, N. Lee*, T. Kang*, T. Kim, D. Kim, D. Lee, S. Oh, W. Gong, I. Y. Kwak
“Auxiliary-conditioned Cross-Attention with Physiologically Interpretable Features for Chagas Disease Detection from 12-lead ECGs.”
Physiological Measurement, 2026.
* Equal contribution
H. Im, N. Lee, T. Kang, T. Kim, D. Kim, D. Lee, S. Oh, W. Gong, I. Y. Kwak
“ResNet-BiGRU with Conditioned Query-Based Cross-Attention and Weighted Loss for Automated Chagas Disease Detection from 12-Lead ECG.”
Computing in Cardiology (CinC 2025), 2025.
- KD(Knowledge Distillation) Basic [바로가기]
- Paper Follow Up - 26.08 5주차 [바로가기]
- Paper Follow Up - 26.08.4주차 [바로가기]
- SAE(Sparse AutoEncoder), Steering Vector [바로가기]
- tidyvocie2026 [바로가기]
