This book explores the emerging role of performance predictors in evolutionary neural architecture search (ENAS), highlighting representative methods and practical applications that make this field both timely and impactful. By bridging performance prediction with evolutionary optimization, it addresses one of the most pressing challenges in deep learning: how to efficiently design and evaluate neural architectures without incurring prohibitive computational costs.The book provides a systematic overview of predictor-driven approaches across diverse neural network model families, including graph neural networks, convolutional neural networks, and fuzzy neural networks. It introduces accuracy predictors as well as rank-aware predictors, illustrating how these methods enhance the efficiency, scalability, and generalizability of neural architecture search. In addition to results on widely used benchmark datasets, the book emphasizes practical applications such as defect detection and medical image segmentation, showcasing how predictor-guided ENAS delivers both research insights and real-world impact.By engaging with this book, readers will gain a clear understanding of how performance predictors accelerate ENAS, discover both classical techniques and recent advances, and appreciate the methodological and applied value of predictor-guided architecture design. The book equips its audience with frameworks to evaluate and extend predictor-based methods, positioning them at the intersection of evolutionary computation, performance prediction, and neural architecture search. Additionally, the code related to the book will be available as open source.This volume is intended for researchers, graduate students, and professionals seeking to deepen their expertise in evolutionary computation, neural networks, and neural architecture search. A foundational background in these areas will facilitate full engagement with the material and enable readers to leverage the presented concepts for both academic inquiry and applied innovation.

This book explores the emerging role of performance predictors in evolutionary neural architecture search (ENAS), highlighting representative methods and practical applications that make this field both timely and impactful. By bridging performance prediction with evolutionary optimization, it addresses one of the most pressing challenges in deep learning: how to efficiently design and evaluate neural architectures without incurring prohibitive computational costs.The book provides a systematic overview of predictor-driven approaches across diverse neural network model families, including graph neural networks, convolutional neural networks, and fuzzy neural networks. It introduces accuracy predictors as well as rank-aware predictors, illustrating how these methods enhance the efficiency, scalability, and generalizability of neural architecture search. In addition to results on widely used benchmark datasets, the book emphasizes practical applications such as defect detection and medical image segmentation, showcasing how predictor-guided ENAS delivers both research insights and real-world impact.By engaging with this book, readers will gain a clear understanding of how performance predictors accelerate ENAS, discover both classical techniques and recent advances, and appreciate the methodological and applied value of predictor-guided architecture design. The book equips its audience with frameworks to evaluate and extend predictor-based methods, positioning them at the intersection of evolutionary computation, performance prediction, and neural architecture search. Additionally, the code related to the book will be available as open source.This volume is intended for researchers, graduate students, and professionals seeking to deepen their expertise in evolutionary computation, neural networks, and neural architecture search. A foundational background in these areas will facilitate full engagement with the material and enable readers to leverage the presented concepts for both academic inquiry and applied innovation.