Picture

​Chia-Yi Hsu
[email protected]
​​PhD student at National Yang Ming Chiao Tung University, Taiwan.
Research Interest
Safety of Large Language/Diffusion Models, Robustness of Neural Networks, Adversarial Learning, (Local) Differential Privacy.
Journal Publiactions
  1. Yan-Ting Chen, Chia-Yi Hsu, Chia-Mu Yu, Mahmoud Barhamgi, Charith Perera. On The Private Data Synthesis Through Deep Generative Models for Data Scarcity of Industrial Internet of Things. IEEE Transactions on Industrial Informatics.
  2. Chih-Hsun Lin, Chia-Yi Hsu, Chia-Mu Yu, Yang Cao, Chun-Ying Huang. DPAF: Image Synthesis via Differentially Private Aggregation in Forward Phase. IEEE Internet of Things Journal.
Conference Publications
  1. Yan-Lun Chen, Yi-Ru Wei, Chia-Yi Hsu, Chia-Mu Yu, Chun-Ying Huang, Ying-Dar Lin, Yu-Sung Wu, Wei-Bin Lee. Layer-Aware Task Arithmetic: Disentangling Task-Specific and Instruction-Following Knowledge. The Conference on Empirical Methods in Natural Language Processing, (EMNLP), 2025.
  2. Chia-Yi Hsu, Jia-You Chen, Yu-Lin Tsai, Chih-Hsun Lin, Pin-Yu Chen, Chia-Mu Yu, Chung-Ying Huang. VP-NTK: Exploring the Benefits of Visual Prompting in Differentially Private Data Synthesis. IEEE International Conference on Acoustics, Speech and Signal Processing, (ICASSP), 2025​.
  3. Chia-Yi Hsu, Yu-Lin Tsai, Chih-Hsun Lin, Pin-Yu Chen, Chia-Mu Yu, Chung-Ying Huang. SafeLoRA: The Silver Lining of Reducing Safety Risks when Finetuning Large Language Models. The Thirty-Eighth Annual Conference on Neural Information Processing Systems, (NeurIPS), 2024.
  4. Wei-Jia Chen, Chia-Yi Hsu, Wei-Bin Lee, Chia-Mu Yu, Chung-Ying Huang. Road Decals as Trojans: Disrupting Autonomous Vehicle Navigation with Adversarial Patterns.  The 54th Annual IEEE/IFIP International Conference on Dependable Systems and Networks (DSN), 2024.
  5. Yu-Lin Tsai*, Chia-Yi Hsu*, Chulin Xie, Chih-Hsun Lin, Jia You Chen, Bo Li, Pin-Yu Chen, Chia-Mu Yu, Chun-Ying Huang. Ring-A-Bell! How Reliable are Concept Removal Methods For Diffusion Models? The Twelfth International Conference on Learning Representations (ICLR), 2024.  * equal contribution.
  6. Hsiu-Fu Wu, Chia-Yi Hsu, Chih-Hsun Lin, Chia-Mu Yu, Chun-Ying Huang. Deepfake Detection through Temporal Attention. The 33rd Wireless and Optical Communications Conference (WOCC), 2024.
  7. Chang-Sheng Lin, Chia-Yi Hsu, Pin-Yu Chen, Chia-Mu Yu. Real-World Adversarial Examples involving Makeup Application. IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2022.
  8. Chia-Yi Hsu, Pin-Yu Chen, Songtao Lu, Sijia Liu, and Chia-Mu Yu. Adversarial Examples can be Effective Data Augmentation for Unsupervised Machine Learning. AAAI Conference on Artificial Intelligence (AAAI), 2022.
  9. Yu-Lin Tsai, Chia-Yi Hsu, Pin-Yu Chen, and Chia-Mu Yu. Formalizing Generalization and Robustness of Neural Networks to Weight Perturbations. Conference on Neural Information Processing Systems (NeurIPS), 2021.
  10. Xiao Jin, Pin-Yu Chen, Chia-Yi Hsu, Chia-Mu Yu, Tianyi Chen. Catastrophic Data Leakage in Vertical Federated Learning. Conference on Neural Information Processing Systems (NeurIPS), 2021.​
  11. Yu-Lin Tsai, Chia-Yi Hsu, Pin-Yu Chen and Chia-Mu Yu. Non-Singular Adversarial Robustness of Neural Networks. IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2021.
  12. Chia-Yi Hsu, Pin-Yu Chen, and Chia-Mu Yu. Towards Information Theoretic Adversarial Examples. Workshop on Secure and Resilient Autonomy (SARA), in conjunction with The Conference on Machine Learning and Systems (MLSys), March 4, Austin, Texas, U.S.A., 2020.
  13. Chia-Yi Hsu, Pin-Yu Chen, Chia-Mu Yu. Characterizing Adversarial Subspaces by Mutual Information. ACM ASIA Conference on Computer and Communications Security (ASIACCS), 2019. (Poster)
  14. Chia-Yi Hsu, Pei-Hsuan Lu, Pin-Yu Chen, Chia-Mu Yu. On The Utility of Conditional Generation Based Mutual Information for Characterizing Adversarial Subspaces. IEEE Global Conference on Signal and Information Processing (GlobalSIP), 2019.​
Patent
  • Generating unsupervised adversarial examples for machine learning.  
  • Neural Network Robustness through obfuscation. 
Honor
Chi-Sung Laih Memorial Master Thesis Award
Chinese Cryptology and Information Security Association, 2021.
ACM ASIACCS Conference
  • Travel grant for ACM ASIACCS 2019.
Ministry of Science and Technology Scholarship
  • Travel grant for GlobalSIP 2019.
Research Experience
Institue of Science Tokyo (2025/11 - now)
Visiting Scholar
Advisor: Jun Sakuma
CISPA Helmholtz Center for Information Security (2024/12 - 2025/10, 10 months)
Visiting Scholar
Advisor: Yang Zhang
CISPA Helmholtz Center for Information Security (2024/07 - 2024/09, 3 months)
Visiting Scholar
Advisor: Yang Zhang
  • Unveiling Backdoor Threats in Third-Party Task Vectors​
Thomas J. Watson Research Center IBM, USA (2021/07 – 2021/08, 2 months)                                                                                            
Visiting Scholar
Advisor: Pin-Yu Chen
  • Data Leakage in Federated Learning
  • Weight Perturbation of Neural Networks
Thomas J. Watson Research Center IBM, USA (2019/08 – 2020/08, 1 year)                                                                                          
Visiting Scholar
Advisors: Pin-Yu Chen
  • Unsupervised Adversarial Examples
  • Obfuscate Gradients with The Sine Function
Technische Universität Darmstadt (TUD), Germany (2019/01 – 2019/02, 1 month)                                                                                                 
Visiting Student
Advisors: Neeraj Suri
  • Software Testing Tools (cdf and klee)
Center for Secure Energy Informatics (CSE), FH Salzburg, Austria (2018/01 – 2018/02, 1 month)                                                                                                
Visiting Student
Advisor: Dominik Engel
  • Design of local differentially private smart metering