Recent Advances in Trustworthy Deep Learning on Images

"Trustworthiness has been a huge challenge for deep learning. In this talk, I will cover this topic from different perspectives, including adversarial robustness, out-of-distribution detection, and model interpretability. I will give a comprehensive introduction of our recent works that are done in the past three years in my research lab (Robust and Interpretable Machine Learning Lab), and conclude the talk with future research directions.

"