This book masters the critical skill of selective data removal from AI systemsessential for regulatory compliance and ethical AI development. This comprehensive book bridges the gap between theoretical foundations and practical implementation, offering clear pathways through both exact and approximate unlearning methodologies. Designed for machine learning engineers, privacy specialists, researchers, and policymakers, it uniquely integrates technical depth with legal and ethical frameworks. From telecom to finance, discover how to eliminate data influence while preserving model utility. Ideal for graduate courses, professional training, and organizational compliance initiatives, this book positions you at the forefront of responsible AI innovation in an increasingly privacy-conscious world.

Provides a systematic exploration of machine unlearning's conceptual frameworks Presents core principles and methods to unlearn data via exact and approximate strategies through parameter adjustments Explores unlearning protocols for various architectures, from transformers to GNNs and recommendation systems

Inhalt

Chapter 1: Introduction to Machine Unlearning.- Chapter 2: Exact Unlearning Methods.- Chapter 3: Approximate Unlearning Technique.- Chapter 4: Machine Unlearning for Different Model Architectures.- Chapter 5: Verification and Evaluation of Unlearning.- Chapter 6: Applications of Machine Unlearning.- Chapter 7: Ethics, Privacy, Security, Governance, and Future Considerations.