Evaluation of Unlearning and Graph Unlearning - Part 2

This lecture explores the evaluation methodologies for machine unlearning and specifically graph unlearning techniques. Dive into key evaluation frameworks including WMDP (Weak Membership Decision Problem) and TOFU (Test Of Forgetting Utility), while examining benchmark datasets used to assess unlearning effectiveness. The 21-minute presentation from NPTEL-NOC IITM provides detailed insights into how unlearning processes are measured and validated in graph-based models, with supplementary lecture materials available for download to enhance understanding of these complex concepts.