user_image

Daeyoung Kim

I am a Research Engineer at NCSoft. I received my M.S from KAIST Graduate school of AI, where I was advised by Edward Choi.

RESEARCH INTERESTS

  • Multi-modality in AI
  • Reliability of Machine Learning
  • Machine Learning for Healthcare

EXPERIENCE

  • 2023-02 – Present

    NCSoft

    Large Multimodal Model

    Alignment Tuning (RLHF, RLAIF)

    LLM-as-an-evaluator

    Writing Assistant using LLM

  • 2022-07 – 2023-01

    NAVER CLOVA, Research Engineer Intern

    Sentence Embedding using Large Language Model

    Zero-shot Text Classification

  • 2020-07 – 2020-08

    NAVER Corp, Research Engineer Intern

    Sentimental Analysis of Short Sentence

    Data Augmentation

  • 2020-01 – 2020-12

    Data Mining and Information Systems Lab, Undergraduate Researcher

    Zero-shot Text Summarization

EDUCATION

  • 2021-03 – 2023-02

    Korea Advanced Institute of Science and Technology

    Master of Artificial Intelligence (Graduate School of AI)

    Advisor Prof. Edward Choi

  • 2015-03 – 2021-02

    Korea University

    Bachelor of Department of Computer Science and Engineering

    Bachelor of Integrated Major in Information Security Convergence

PUBLICATIONS

  1. OffsetBias: Leveraging Debiased Data for Tuning Evaluators
    Junsoo Park*, Seungyeon Jwa*, Meiying Ren,  Daeyoung Kim, and Sanghyuk Choi
    In Findings in Empirical Methods in Natural Language Processing, (EMNLP), 2024
  2. Towards the Practical Utility of Federated Learning in the Medical Domain
    Seongjun Yang, Hyeonji Hwang,  Daeyoung Kim, Radhika Dua, Jong-Yeup Kim, Eunho Yang, and Edward Choi
    In Proc. of Conference on Health, Inference, and Learning, (CHIL), 2023
  3. Revisiting the Importance of Amplifying Bias for Debiasing
    Jungsoo Lee*, Jeonghoon Park*,  Daeyoung Kim*, Juyoung Lee, Edward Choi, and Jaegul Choo
    In Proc. of Association for the Advancement of Artificial Intelligence, (AAAI), 2023, Accepted as Oral Presentation
  4. Uncertainty-Aware Text-to-Program for Question Answering on Structured Electronic Health Records
    Daeyoung Kim, Seongsu Bae, Seungho Kim, and Edward Choi
    In Proc. of Conference on Health, Inference, and Learning, (CHIL), 2022
  5. Question Answering for Complex Electronic Health Records Database using Unified Encoder-Decoder Architecture
    Seongsu Bae,  Daeyoung Kim, Jiho Kim, and Edward Choi
    In Proc. of Machine Learning for Health, (ML4H), 2021, Accepted as Oral Presentation
  6. TeSS: Zero-Shot Classification via Textual Similarity Comparison with Prompting using Sentence Encoder
    Jimin Hong*, Jungsoo Park*,  Daeyoung Kim*, Seongjae Choi, Bokyung Son, and Jaewook Kang
    Under Review

* denotes equal contribution.

AWARDS

  • Capstone Design at Korea Univ. (3rd Place)
  • Text Summarization of Biomedical paper related to COVID-19. [GitHub]
  • HeLP Challenge 2019 at Asan Medical Center (1st Place)
  • Breast Cancer Classification on Frozen Pathology. [GitHub]
  • Kakao Arena (3rd Place)
  • Article Recommendation Task. [GitHub]