Strategic Feature Selection
Researchers have formally studied strategic classification through feature selection and its interaction with ridge regularization. Their findings indicate that excluding individual features based solely on manipulability is often suboptimal. The study proposes a practical algorithm for jointly selecting feature sets and ridge regularization levels, offering a framework to mitigate strategic behavior in algorithmic decision-making systems, particularly in high-stakes domains like healthcare. AI
IMPACT Provides a principled framework for mitigating strategic behavior in algorithmic decision-making systems, applicable to high-stakes domains like healthcare.