Please join the Department of Biomedical Informatics and Data Science (DBIDS) for the upcoming edition of the PowerTalks Seminar Series, featuring Dongxiao Zhu, Ph.D., professor of Computer Science and the founding director of the Wayne AI Research Initiative at Wayne State University. 

Dr. Zhu's discussion, entitled "Enhancing Algorithmic Fairness in Deep Learning: Discovering and Mitigating Spurious Features," will introduce a recent approach to enhancing algorithmic fairness in DNN by identifying and mitigating spurious patterns. The quality of the explanations will be evaluated quantitatively and qualitatively using imaging and text datasets from Convolutional Neural Network (CNN) and Transformer architectures. The presentation will also discuss recent work on fairness-through-attack for Vision Transformer (ViT), which involves learning spurious features that are exclusively correlated with sensitive attributes (such as gender) and using adversarial attacks to eliminate the spurious patterns through debiased self-attention. 

In 2022, Dr. Zhu was awarded the Excellence in Research Award for Senior Faculty from the College of Engineering at Wayne State University. Over the past five years, his research has been supported by five NSF grants, two NIH grants, and private foundations, totaling over 6 million dollars