Skip to content
View BBongjun's full-sized avatar
  • Seoul National University of Science and Technology
  • Seoul, South Korea

Highlights

  • Pro

Block or report BBongjun

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.

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
BBongjun/README.md

안녕하세요, 최봉준입니다 👋

👤 About Me

  • M.S. in Data Science, Seoul National University of Science and Technology
  • B.S. in Industrial & Information Systems Engineering, Seoul National University of Science and Technology
  • Research focus: Data Mining, Deep Learning, Time Series Analysis
  • Building models that do not stop at anomaly detection, but also explain why they were judged anomalous

🔬 Research

Prototyping Inter-variable Relationships for Explainable Multivariate Time-Series Anomaly Detection

  • M.S. Thesis
  • Built a method that extracts inter-variable interactions with an Inverted Transformer
  • Clustered normal attention maps with FINCH to build normal relationship prototypes
  • Combined reconstruction score and prototype discrepancy score for anomaly detection
  • Extended detection to interpretation at the time / variable / relation level

Multiple Instance Learning for SSD Failure Prediction under Customer Failure-biased Labels

  • Under 1st major revision, Computers & Industrial Engineering
  • Addressed noisy labels in SSD failure logs
  • Grouped suspicious SSD sequences into bags and trained with Multiple Instance Learning
  • Reduced the impact of inaccurate failure labels and improved precision

🏆 Competition

ETRI Fashion-How Season 5

  • Worked on model compression and zero-shot recommendation tasks
  • Improved lightweight classification performance with Knowledge Distillation
  • Improved Top-1 Accuracy from 35.2% → 42.4%
  • Improved recommendation metric WKTC from 0.3294 → 0.5557

KAMP Competition | Real-time Defect Detection under Concept Drift

  • Proposed an AI framework for manufacturing environments with Concept Drift
  • Applied D3 drift detection + Active Learning + online model update
  • Reduced labeling cost by over 60%
  • Maintained stable defect detection performance in streaming environments

🎖 Awards

  • Outstanding Master’s Thesis Award Prototyping Inter-variable Relationships for Explainable Multivariate Time-Series Anomaly Detection
  • Encouragement Award, 2024 ETRI FASHION-HOW Competition
  • Silver Prize, Capstone Design Presentation
    XAI-based Study on Fake Reviews Generated by ChatGPT

🎓 Education

  • Seoul National University of Science and Technology M.S. in Data Science (2024.03 ~ 2026.02)
    GPA: 4.27 / 4.5
  • Seoul National University of Science and Technology B.S. in Industrial & Information Systems Engineering (2018.03 ~ 2024.02)
    GPA: 4.02 / 4.5

🛠 Tech Stack

Languages & Frameworks

Python PyTorch Scikit-learn

Data & Analysis

Pandas NumPy Matplotlib Jupyter

Tools

Git GitHub

Interests

Time Series Anomaly Detection Explainable AI Industrial AI


📚 Selected Coursework

  • Deep Learning
  • Data Mining
  • Advanced Machine Learning
  • Explainable Machine Learning
  • Operations Research
  • Python Programming
  • Engineering Statistics
  • Linear Algebra

✉️ Contact

Pinned Loading

  1. Self-Filtering_reproduction Self-Filtering_reproduction Public

    Forked from 1998v7/Self-Filtering

    Reproduction study about ECCV 2022 Paper “Self-Filtering: A Noise-Aware Sample Selection for Label Noise with Confidence Penalization”

    Jupyter Notebook

  2. 2024_KAMP 2024_KAMP Public

    Jupyter Notebook

  3. DataMining_project DataMining_project Public

    데이터마이닝 팀프로젝트 1조 최종 자료

    Jupyter Notebook 1

  4. DataBase-Project DataBase-Project Public

    2021학년도 2학기 데이터베이스 팀프로젝트

    Jupyter Notebook

  5. RTM_Drystrip_GDN RTM_Drystrip_GDN Public

    Application of GDN to DryStrip Data

    Python

  6. passion3659/ssd-failure passion3659/ssd-failure Public

    Jupyter Notebook 1 1