This book explores the power of genetic programming for symbolic regression (GPSR), offering a unique pathway to discovering models that are both highly accurate and naturally interpretable. With a clear focus on the twin goals of generalisation and interpretability, the book introduces cutting-edge techniques such as representation learning, complexity control, semantic-aware operators, and multi-objective optimisation for GPSR. Each chapter blends foundational theory with empirical investigation, guiding readers through model development, evaluation, and experiments on a variety of scientific and engineering datasets.

A central theme of this book is generalisationthe ability of a model to perform well on new, unseen data. Unlike traditional symbolic regression methods that risk overfitting, the techniques presented in this book aim to evolve models that not only fit the training data but also maintain performance on new, independent data. Readers will explore principled approaches to regularisation, semantic diversity preservation, and fitness evaluations to promote generalisation, all designed to build models that are robust and reliable in practical settings.

Designed for researchers, data scientists, and research students, this book provides practical tools to evolve symbolic models that are interpretable, trustworthy, and effective in capturing meaningful patterns in data. Readers will benefit from structured frameworks for building interpretable models, proven strategies to reduce overfitting and improve robustness, and insights into model interpretability. Engaging case studies and examples throughout the book bring these methods to life, making Evolving Insights an essential resource for anyone seeking clarity and trust in machine learning.

Discusses both theoretical and methodological improvements of genetic programing for symbolic regression (GPSR) Features generalization and interpretation of GPSR as a good integration of GP and modern machine learning Provides case studies and real-world application problems to generate both academic impact and real-world impact

Autorentext

Qi Chen is a Senior Lecturer in Artificial Intelligence at Victoria University of Wellington, New Zealand. Her research expertise lies in genetic programming, symbolic regression, evolutionary computation, and explainable AI, with a strong emphasis on model interpretability and generalization. Dr. Chen has published extensively in top-tier journals, including IEEE Transactions on Pattern Analysis and Machine Intelligence, IEEE Transactions on Evolutionary Computation and IEEE Transactions on Cybernetics, ECJ and GPEM, and has presented her work at major conferences such as GECCO, PPSN, CEC and Evo Star. She is an active reviewer and program committee member for leading venues in evolutionary computation and machine learning. Dr. Chen also leads research projects supporting PhD.

Bing Xue is a Professor of Artificial Intelligence at Victoria University of Wellington, New Zealand, a Fellow of IEEE, and a Fellow of Engineering New Zealand. Her research focuses on evolutionary computation, symbolic regression, feature selection, multi-objective optimization, and automated machine learning. She has authored over 500 peer-reviewed publications, including in top-tier journals such as IEEE Transactions on Evolutionary Computation (TEVC), IEEE Transactions on Cybernetics (TCYB), ECJ, and Information Sciences. Professor Xue is an Associate Editor for leading journals such as IEEE TEVC and ACM Transactions on Evolutionary Learning and Optimization, and regularly serves as Conference/Program Chair or Senior PC Member for major international conferences, including EuroGP, ACM GECCO, IEEE CEC, and PRICAI. She is also a Clarivate Reveals Highly Cited Researcher 2023 and 2024.

Mengjie Zhang is a Professor of Computer Science at Victoria University of Wellington, New Zealand. He is a Fellow of the Royal Society of New Zealand, a Fellow of IEEE, and a Fellow of Engineering New Zealand. Professor Zhang is internationally recognized for his influential work in genetic programming, evolutionary computation, evolutionary machine learning, symbolic regression, feature selection, and image analysis. He has authored over 800 peer-reviewed publications in top venues such as IEEE Transactions on Evolutionary Computation, IEEE Transactions on Cybernetics, and GECCO. He received the “Evo* Award for Outstanding Contribution to Evolutionary Computation in Europe 2023” and the “2024 Australasian Artificial Intelligence Distinguished Research Contribution Award”. He is also a Clarivate Reveals Highly Cited Researcher 2023 and 2024.

Inhalt

"Part I Fundamentals and Motivations".- "1.Introduction to Symbolic Regression".- "2.Genetic Programming for Symbolic Regression".- "3 Generalisation and Interpretability".- "4 Benchmarking and Open Tools in GPSR".- "Part II Techniques and Innovations Driving Generalisation and Interpretability in GPSR".- "5.Feature Selection for GPSR".- "6.Controlling Complexity via Structural Risk Minimisation in GPSR".- "7.Angle Distance Driven Geometric Semantic GPSR".- "8.Semantic Transformations for Population Diversity Preservation in
GPSR".- "Part III Emerging Techniques and Applications".- "9.Transfer Learning and Multi-task GPSR".- "10.GPSR with Hybrid Models".- "11.Applications and Real-World Impact of GPSR".- "Part IV.Conclusions and Open Directions".- "12 Summary of Key Insights in GPSR".