Tuning: Methodology

Thomas Bartz-Beielstein, Martin Zaefferer, Olaf Mersmann

Abstract: This chapter lays the groundwork and presents an introduction to the process of tuning Machine Learning (ML) and Deep Learning (DL) hyperparameters and the respective methodology used in this book. The key elements such as the hyperparameter tuning process and measures of tunability and performance are defined. Practical considerations are presented and all the ingredients needed for successful hyperparameter tuning are explained. A special focus lies on how to prepare the data. This might be the most thorough overview presented yet.