Fundamentals of Robust Machine Learning

Bechoo Lal

• Publisher: Toronto Academic Press
• eBook ISBN:
  9781834412467
• Year: 2026

The reliability and stability of machine learning systems are crucial in real-world applications, where data variability, noise, and uncertainty can significantly affect model performance. Building models that remain effective under such challenges defines the essence of robust AI. Fundamentals of Robust Machine Learning explores the theoretical and practical approaches to designing resilient learning algorithms. The book discusses adversarial robustness, data augmentation, uncertainty quantification, and generalization techniques. It also covers robust optimization and fairness in model evaluation. Combining mathematical rigor with applied examples, it provides students, researchers, and engineers with tools to build dependable machine learning systems capable of handling complex, imperfect, and evolving data environments.

Table of contents

• Cover
• Title Page
• Copyright
• About the Author
• Brief Contents
• Getting Started with This Book
• Table of Contents
• List of Figures
• Preface
• CHAPTER 1: INTRODUCTION TO ROBUST MACHINE LEARNING
• CHAPTER 2: THREAT MODELS IN MACHINE LEARNING
• CHAPTER 3: BASICS OF ADVERSARIAL ATTACKS
• CHAPTER 4: DEFENSES AGAINST ADVERSARIAL ATTACKS
• CHAPTER 6: CERTIFIED DEFENSES
• CHAPTER 7: ROBUSTNESS TO DATA POISONING
• CHAPTER 8: OUT-OF-DISTRIBUTION (OOD) GENERALIZATION
• CHAPTER 9: GROUP ROBUSTNESS AND FAIRNESS
• CHAPTER 10: ROBUSTNESS IN NATURAL LANGUAGE PROCESSING
• CHAPTER 11: ROBUSTNESS IN COMPUTER VISION
• CHAPTER 12: EVALUATION AND BENCHMARKING OF ROBUST MACHINE LEARNING MODELS
• CHAPTER 13: FRONTIERS AND OPEN PROBLEMS IN ROBUST ML
• INDEX
• Back Cover