Strategic Classification

Strategic classification studies machine learning models' robustness against strategic manipulation by users aiming to receive favorable predictions. Current research focuses on developing algorithms and models, including those based on minimax fairness, online learning frameworks (like the strategic perceptron), and graph neural networks, that are resilient to such manipulations, often considering various levels of information availability and agent cost structures. This field is significant because it addresses the crucial issue of model reliability and fairness in real-world applications where users can actively influence the input data, impacting areas like loan applications, college admissions, and social network analysis.