Skip to content
View Laplace-tech's full-sized avatar

Block or report Laplace-tech

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
Laplace-Tech/README.md

My Neural Cloud Hub

Profile

Profile Details
Name Yongmin Park (박용민)
Academic Affiliation 경기대학교(수원) · 컴퓨터공학전공 (Entered in 2022)
Academic Status 3rd-Year Undergraduate Student
Research Interests Medical AI · Medical Imaging · AI for Healthcare
Email add28482848@kyonggi.ac.kr

Current Research

3D 복부 CT Segmentation에서 organ-wise learning states와 error types을 계층적 조건부 확률 구조로 모델링한 Adaptive Patch Sampling 연구.

  • TotalSegmentator v2.0.1과 nnU-Net v2.8.1 3d_fullre를 baseline으로 사용하고, 동일한 학습 조건에서 Patch Sampling Policy만 단계적으로 변경하여 adaptive sampling의 효과를 정량적으로 비교
  • 2026 한국정보기술학회(KIIT) 추계종합학술대회 대학생논문경진대회 참가 (On-going)

Featured Work

전통적인 CNN 기반의 Chest X-ray classification model과 Grad-CAM 시각화 기법을 web application으로 통합한 의료영상 판독 보조 프로토타입.

  • Multi-label classification과 Grad-CAM을 결합한 AI architecture 설계
  • Data preparation, model training, inference, service integration을 잇는 end-to-end R&D pipeline 구축
  • 2026 경기대학교 산학협력 캡스톤디자인 경진대회 기초캡스톤 부문 장려상 수상

Stanford ML Group이 공개한 CheXpert 데이터셋을 활용해 multi-label chest X-ray classification 모델의 학습·평가 파이프라인을 재현한 PoC.

  • Uncertainty label policy 비교, class-wise evaluation, threshold tuning과 Grad-CAM 기반 error analysis를 포함한 reproducible experiment pipeline 구축
  • 2026 한국정보기술학회(KIIT) 하계종합학술대회 대학생 부문 논문경진대회 우수논문상 은상 수상


Technical Stack

AI & Experimentation

Python PyTorch NumPy pandas scikit-learn Jupyter

Backend & Data

FastAPI Java Spring Boot PostgreSQL

Infrastructure & Development

Docker Linux Git GitHub VS Code

Footer

Pinned Loading

  1. capstone-cxr capstone-cxr Public

    MediScope: end-to-end chest X-ray reading assistance prototype integrating DenseNet121 classification, Grad-CAM explainability, and a full-stack web service.

    Python 1 2

  2. hierarchical-patch-sampling-3d hierarchical-patch-sampling-3d Public

    Hierarchical conditional-probability-based patch sampling for 3D abdominal CT multi-organ segmentation. Manuscript, results, figures, and study notebooks.

    Jupyter Notebook 1

  3. CheXpert CheXpert Public

    CheXpert-based chest X-ray multi-label classification research PoC with DenseNet121, uncertainty-label experiments, threshold tuning, and Grad-CAM explainability.

    Python 1

  4. maverick maverick Public

    PyTorch implementations and executable study notes for Dive into Deep Learning (D2L).

    Jupyter Notebook 1