ERASURE: Redefining Privacy Through Selective Machine Unlearning

ERASURE: Redefining Privacy Through Selective Machine Unlearning

Overview

ERASURE offers fully extensible built-in components, allowing users to define custom unlearning techniques, integrate custom and synthetic datasets, implement tailored evaluation metrics, and seamlessly work with state-of-the-art machine learning models.

Machine Unlearning (MU) enables selective removal of learned information from AI models without full retraining, addressing critical needs in:

  • Data Privacy (GDPR Right to be Forgotten)
  • Bias Mitigation
  • Regulatory Compliance (AI Act, CCPA)

Resources

  • The Forget-Set Identification Problem – Machine Learning (Springer), 2025
  • WIPE-OUT Workshop – ECML-PKDD 2025 & 2026