Joohyung Lee

Contact: joohyunglee [dot] research [at] gmail [dot] com

prof_pic.jpg

Machine Learning Researcher

at AITRICS

I am a machine learning researcher at AITRICS, a healthcare AI startup in South Korea, advised by Prof. Eunho Yang and collaborating closely with Prof. Juho Lee.

My research focuses on inductive bias and symmetry-aware learning: how to design learning systems that generalize reliably by encoding principled regularities (e.g., geometry, acquisition processes, and temporal structure). I have worked on geometric priors [1], acquisition-informed priors [2, 3], and temporal/observation-process priors for multimodal EHR trajectories [4, 5].

Previously, I worked as a researcher at the Korea Electronics Technology Institute (KETI) and in the RCV Lab at KAIST with Prof. In So Kweon. I also collaborated with clinicians and researchers at the National Cancer Center.

I am currently investigating invariance-equivariance interference. During my PhD, I aim to move beyond hand-specified groups and geometry: toward learned symmetries and semantic/physical plausibility constraints (Research Statement).

News

Mar 02, 2026 Two papers, “Status-Aware Self-Supervised Forecasting for Irregular Clinical Time Series” and “Structure-Aware Set Transformers: Temporal and Variable-type Attention Biases for Asynchronous Clinical Time Series,” have been accepted to the ICLR 2026 Time Series in the Age of Large Models Workshop.
Jan 26, 2026 One paper Soft Equivariance Regularization for Invariant Self-Supervised Learning was accepted by ICLR 2026.

Selected Publications

  1. ICLR Workshop
    Structure-Aware Set Transformers: Temporal and Variable-type Attention Biases for Asynchronouse Clinical Time Series
    Joohyung Lee, Kwanhyung Lee, Changhun Kim, and 1 more author
    In 1st ICLR Workshop on Time Series in the Age of Large Models, 2026
    Accepted
  2. ICLR Workshop
    Status-Aware Self-Supervised Forecasting for Irregular Clinical Time Series
    Kwanhyung Lee, Joohyung Lee, Jong-Heon Kim, and 2 more authors
    In 1st ICLR Workshop on Time Series in the Age of Large Models, 2026
    Accepted
  3. Soft Equivariance Regularization for Invariant Self-Supervised Learning
    Joohyung Lee, Changhun Kim, Hyunsu Kim, and 2 more authors
    In International Conference on Learning Representations (ICLR), 2026
    Accepted
  4. Compact and De-Biased Negative Instance Embedding for Multi-Instance Learning on Whole-Slide Image Classification
    Joohyung Lee, Heejeong Nam, Kwanhyung Lee, and 1 more author
    In IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2024
  5. Learning Missing Modal Electronic Health Records with Unified Multi-Modal Data Embedding and Modality-Aware Attention
    Kwanhyung Lee, Soojeong Lee, Sangchul Hahn, and 4 more authors
    In Machine Learning for Health (MLHC), 2023
    Spotlight; Corresponding Author and Co-first Author
  6. ICLR Workshop
    Self-Supervised Predictive Coding with Multimodal Fusion for Patient Deterioration Prediction in Fine-Grained Time Resolution
    Kwanhyung Lee, John Won, Heejung Hyun, and 3 more authors
    In ICLR 2023 Trustworthy Machine Learning for Healthcare Workshop, 2023
    Oral Presentation; Corresponding Author; Best Paper Honorable Mention
  7. Moving from 2D to 3D: Volumetric Medical Image Classification for Rectal Cancer Staging
    Joohyung Lee, Jieun Oh, Inkyu Shin, and 4 more authors
    In International Conference on Medical Image Computing and Computer-Assisted Intervention (MICCAI), 2022
  8. Reducing the Model Variance of Rectal Cancer Segmentation Network
    Joohyung Lee, Ji Eun Oh, Min Ju Kim, and 2 more authors
    IEEE Access, 2019