Siyi Chen awarded Predoctoral Fellowship to support research on Controllable and Interpretable AI models

Siyi Chen, Ph.D. student in Electrical and Computer Engineering, was awarded a Rackham Predoctoral Fellowship to support her research in generative and perceptive AI models, with the goal of enhancing the usability, safety, and reliability of AI systems in real-world applications. 

Her proposed dissertation title is “Advancing Interpretability and Controllability in AI-powered Generation and Perception.”

“AI systems play an increasingly prominent role in public safety monitoring, healthcare, and content creation,” explains Chen. “However, the opaque, complex nature of deep neural networks limits user control, raises ethical concerns, and reduces trust in their outputs, particularly in sensitive applications such as data generation. Hence, the need for interpretable and controllable AI models has become urgent.”

“My dissertation research tackles these issues by advancing interpretable methods to enhance user control, safety, and robustness in AI-powered generative and perceptive models. These advancements equip users with tools to understand and guide AI outputs, promote ethical AI models, and enhance the performance of AI.”

Chen’s dissertation focuses on image generation and image/video learning, specifically: putting more control in the hands of the user; ensuring that harmful or illegal content is not generated; and identifying biases and limitations in existing representation learning models so they can be more appropriate for use in public safety monitoring.

“By developing frameworks to address these challenges, I hope to contribute to the flexible and responsible deployment of AI in society,” said Chen.

Chen received her bachelor’s degree in Electrical and Computer Engineering from the UM-JI joint program with Shanghai Jiao Tong University, and a B.S. in Computer Science from Michigan. She has served as a teaching assistant for four different courses during this time, and served as a reviewer for different professional conferences and journals. She has also served as a mentor to undergraduate and graduate students. Chen is advised by Prof. Qing Qu.

The Rackham Predoctoral Fellowship supports students who are working on dissertations that are “unusually creative, ambitious, and impactful,” and who expect to complete their dissertations during their 3-term fellowship period. Students receive a stipend of more than $40K, tuition, fees, and insurance.