Fairness-Aware Machine Unlearning

Can selectively forgetting training data improve or harm model fairness? This thesis investigates the interplay between machine unlearning and fairness metrics across multiple domains using the ERASURE framework.