Hoki Kim

Professor at Industrial Security, Chuang-Ang University, South Korea.

B.S. and Ph.D. degrees at Seoul National University, South Korea.

Email hokikim.cau@gmail.com

Research Interests. As artificial intelligence (AI) continues to drive innovation across a wide range of industries in the context of the fourth industry revolution, Trustworthy AI (Korean-only) plays a pivotal role in ensuring the reliability and safety of AI systems and mitigating potential risks. As a researcher in machine learning and deep learning, I am currently focused on developing Trustworthy AI with the following topics:

  • Robustness: adversarial attack and defense [NeurIPS 2023, Top AI Conf.; EAAI 2024, IF Top 5%; AAAI 2021, Top AI Conf.; NeuNet, IF Top 10%;]
  • Generalization: domain adaptation [PR, IF Q1], sharpness-aware minimization [ICML 2023, Top AI Conf.]
  • Industrial Applications: audio classification [PR, IF Q1], time series forecasting [ASOC, IF Q1]

Research Experience.

  • Projects with Samsung Electronics, Incheon Airport, and the Ministry of Science and ICT (MSIT), etc.
  • Reviewer in NeurIPS, ICML, IEEE Transactions on Image Processing, and IEEE Transactions on Information Forensics & Security, etc.
  • Developer of torchattacks(★1700+) and torchbnn(★400+).



  1. Fair Sampling in Diffusion Models through Switching Mechanism
    Yujin Choi, Jinseong Park, Hoki Kim, Jaewook Lee, and Saerom Park
    In Proceedings of the AAAI Conference on Artificial Intelligence, 2024
  2. EAAIJournal
    Evaluating practical adversarial robustness of fault diagnosis systems via spectrogram-aware ensemble method
    Hoki Kim, Sangho Lee, Jaewook Lee, Woojin Lee, and Youngdoo Son
    Engineering Applications of Artificial Intelligence, 2024
  1. NeurIPSConference
    Fantastic Robustness Measures: The Secrets of Robust Generalization
    Hoki Kim, Jinseong Park, Yujin Choi, and Jaewook Lee
    In Thirty-seventh Conference on Neural Information Processing Systems, 2023
  2. NeuNetJournal
    Bridged Adversarial Training
    Hoki Kim, Woojin Lee, Sungyoon Lee, and Jaewook Lee
    Neural Networks, 2023
  3. Generating Transferable Adversarial Examples for Speech Classification
    Hoki Kim, Jinseong Park, and Jaewook Lee
    Pattern Recognition, 2023
  4. ASOCJournal
    Fast Sharpness-Aware Training for Periodic Time Series Classification and Forecasting
    Jinseong Park, Hoki Kim, Yujin Choi, Woojin Lee, and Jaewook Lee
    Applied Soft Computing, 2023
  5. Differentially Private Sharpness-Aware Training
    Jinseong Park, Hoki Kim, Yujin Choi, Woojin Lee, and Jaewook Lee
    International Conference on Machine Learning, 2023
  1. TPAMIJournal
    Graddiv: Adversarial Robustness of Randomized Neural Networks via Gradient Diversity Regularization
    Sungyoon Lee, Hoki Kim, and Jaewook Lee
    IEEE Transactions on Pattern Analysis and Machine Intelligence, 2022
  2. Variational Cycle-consistent Imputation Adversarial Networks for General Missing Patterns
    Woojin Lee, Sungyoon Lee, Junyoung Byun, Hoki Kim, and Jaewook Lee
    Pattern Recognition, 2022
  1. Compact Class-conditional Domain Invariant Learning for Multi-class Domain Adaptation
    Woojin Lee, Hoki Kim, and Jaewook Lee
    Pattern Recognition, 2021
  2. Understanding Catastrophic Overfitting in Single-step Adversarial Training
    Hoki Kim, Woojin Lee, and Jaewook Lee
    In Proceedings of the AAAI Conference on Artificial Intelligence, 2021
  1. Torchattacks: A PyTorch Repository for Adversarial Attacks
    Hoki Kim
    arXiv preprint arXiv:2010.01950, 2020