Deletion Inference

Deletion inference explores how information about deleted data can be inferred from machine learning models, before and after data removal, focusing on the implications for privacy and fairness. Current research investigates the robustness of models to deletion attacks using techniques like randomized smoothing and explores the impact of various deletion strategies on model performance and fairness, often employing different machine learning algorithms and datasets. This field is crucial for ensuring compliance with data privacy regulations like GDPR and for developing more robust and ethical AI systems by understanding the vulnerabilities introduced by data deletion and unlearning processes.