Who Am I?

Hi I'm Yonghyeon (Hannah) Kweon and I am pursuing PhD degree in Computer Science. My interest lies on Machine Learning and Deep learning in general, and I am particulary interested in security and privacy concerns in model training of healthcare and medical data.

I received my Master's degree in Data Science an Bachelor's in Statistics at the University of Virginia. Go Hoos!

Education

  •   M.S. in Data Science, 2020
      School of Data Science, University of Virginia

  •   B.A. in Statistics, 2019
      School of Art and Science, University of Virginia

Publications

  • Modeling Route Choice Behavior: A Federated Learning Approach 
    Yonghyeon Kweon, Bingrong Sun, B. Brian Park
    2021 Transportation Research Board (TRB) [Under Review]

    Deep Learning of Protein Structural Classes: Any Evidence for an ‘Urfold’?  [arxiv]
    Menuka Jaiswal, Saad Saleem, Yonghyeon Kweon, Eli J Draizen, Stella Veretnik, Cameron Mura, Philip Bourne
    2020 Systems and Information Engineering Design Symposium (SIEDS)

Research Experience

  • Research Assistant (Jun 2020 - Present) 
    School of Enginneering, University of Virginia

    - Participated in Prof. B. Park's "Route Guidance Recommendation Systme" project
    -Designed Federated Learning model with TensorFlow Fenderated (TFF)
    - Implemented clustering methods based on individual's SVM model to apply federated learning framework

    Research Assistant (Jun 2020 - Present) 
    Mclntire School of Commerce, University of Virginia

    -Prticipated in Prof. Natasha Foutz's "Dynamics in Movie Review" project
    -Scrapped movie data from multiple websites using BeatifulSoup and Selenium
    -Conducted the cleaning/engineering process to investigate the dynamics of movie review with diverse events

    Capstone Research (Jul 2019 - May 2020) 
    School of Data Science, University of Virginia

    -Participated in "Deep Learning for Protein Sturctural Class" project supervised by Prof. Cameron Mura and Prof. Philip Bourne
    -Trained Autoencoder for each protein classification using 3D representation of protein structure
    -Leveraged sparse 3D convolutions to take advantage of data sparsity to make the problem tractable and resource-efficient
    -Produced loss function as a similarity metric between different classes

Relevent Course Projects

  • News Contents Analysis with Natural Language Process (Spring 2020) 
    Conducted Sentimental Analysis and Word Embeddings of news contents using Natural Language Processing (NLP)

    Real Time Face Detection (Spring 2020) 
    Implemented different architectures and pipelines to classify age, gender and emotion of human faces in real time

    Modeling Addmission of Graduate School with Bayesian Inference (Fall 2019) 
    Modeled Bayesian linear regression to predict the uncertainty of the chance of graduate school admission, and utilized hiearchical modeling for candidates' undergraduate school level

    Data Mining for Banknote Authentication (Spring 2019) 
    Used supervised learning models to detect counterfeit banknotes and examined models' performanced-based AUC score