Machine Unlearning for Scientific Discovery

Invited talk on why unlearning matters for scientific workflows (data governance, model updates, and reliability), and how to evaluate it.