Applied Mathematics Seminar——Towards Robust Deep Learning: Risk-Averse Certification and High-Confidence Error Detection
报告人：张喜悦（布里斯托大学计算机学院）
时间：2025-03-19 10:00-12:00
地点：智华楼知无涯-313
报告摘要：
As deep learning systems become increasingly integrated into high-stakes applications, ensuring their reliability and robustness is paramount. This talk explores two complementary techniques for enhancing the reliability of neural networks. First, we introduce RAC-BNN, a Risk-Averse Certification framework for Bayesian neural networks that leverages Conditional Value at Risk (CVaR) to provide probabilistic robustness guarantees under worst-case scenarios. By combining sampling-based uncertainty estimation with optimization techniques, RAC-BNN delivers tighter certified bounds and improved efficiency. Second, we present FAST, a method that enhances deep neural network testing by mitigating the over-confidence problem. FAST dynamically refines uncertainty estimation through guided feature selection, making high-confidence errors more distinguishable and improving the effectiveness of test prioritization. Together, these approaches contribute to building more resilient deep learning models, advancing both certification and testing methodologies for trustworthy AI.
报告人简历:
Xiyue Zhang is a Lecturer in the School of Computer Science at the University of Bristol. Before joining Bristol, she was a Research Associate in the Department of Computer Science at the University of Oxford. She received her PhD in 2022 from Peking University, where she focused on trustworthiness assurance of deep learning systems, and her BSc in 2017, also from Peking University. Her research mainly focuses on trustworthy deep learning, integrating both provable certification and practical testing methods. She is a recipient of the DAAD AInet fellowship 2023 and Future Digileader’23 award by Digital Futures.
