Scholar
Zeyu Qin
Google Scholar ID: 3LXI4-MAAAAJ
Hong Kong University of Science and Technology
Citations & Impact
Citations
451
H-index
10
i10-index
11
Publications
19
Co-authors
15
list available
Publications
Resume (English only)
Academic Achievements
- 2025: Paper 'Scaling Laws of Synthetic Data for Language Models' accepted to COLM 2025
- 2025: Paper 'Safety Reasoning with Guidelines' accepted to ICML 2025, introducing SRG framework for improved out-of-distribution generalization in safety alignment
- 2024: Paper 'Uncovering, Explaining, and Mitigating the Superficial Safety of Backdoor Defense' accepted to NeurIPS 2024 (Spotlight)
- 2023: Two papers accepted to NeurIPS 2023 (one Spotlight), on backdoor purification and imitation learning from imperfect demonstrations
- 2023: First study on robustness from personalization in federated learning against backdoor attacks accepted to KDD 2023
