This book explores one of the most pressing challenges in artificial intelligence: the vulnerability of learning models to adversarial attacks. It presents a clear overview of how such attacks are crafted and the innovative defenses being developed to counter them. Readers will find a balanced mix of theory and practice, with detailed coverage of both text and image domains. Topics range from saliency-based text manipulation and gradient-driven attacks to ensemble frameworks and meta-learning approaches for defense. The volume also highlights hybrid strategies that combine training and preprocessing methods to improve resilience. A distinctive feature of this work is its dual focus: it examines the evolving techniques of attackers while also offering structured and practical defense strategies. Designed for researchers, practitioners, and students, it provides both conceptual clarity and hands-on insights into a rapidly advancing field.
