Abstract: In high-stakes decision settings, individuals may adapt their features to increase their chances of receiving a favorable outcome. Strategic Classification models this phenomenon by assuming that agents face a classifier and incur a cost to modify their features in ways that improve their predicted outcomes. The learner, anticipating such behavior, seeks to design classifiers that remain robust to these strategic adaptations.

However, the standard framework for strategic classification relies on strong and potentially unrealistic assumptions: agents are assumed to have complete knowledge of the deployed classifier; strategic behavior is viewed solely as gaming rather than as potentially self-improving; decisions are modeled as static and offline; and objectives often focus on loss minimization, without explicit regard for fairness or social welfare.

This talk aims to advance our understanding of strategic classification by challenging these modeling assumptions. While I will discuss several directions that relax traditional assumptions, the main focus will be on (i) the role of incomplete information and information asymmetries in strategic classification, (ii) their implications for fairness and social welfare across populations and (iii) the role of causality in distinguishing between ``good’’ and ``bad’’ strategic behavior.

Bio: Juba Ziani is an Assistant Professor in the School of Industrial and Systems Engineering at Georgia Tech and a recipient of the NSF CAREER Award. His research lies at the intersection of computer science, operations research, and economics. He uses tools from learning theory, game theory, and optimization to address technical and societal challenges arising from AI, machine learning, and data-driven decision-making.

Prior to joining Georgia Tech, Juba was a Ph.D. student in Computing and Mathematical Sciences at Caltech, advised by Katrina Ligett and Adam Wierman, and a postdoctoral fellow at the Warren Center for Data Science at the University of Pennsylvania, hosted by Sampath Kannan, Michael Kearns, and Aaron Roth.