Hyperparameter Tuning for Machine and Deep Learning with R: A Practical Guide

This open-access book offers practical examples and insights into hyperparameter tuning for machine learning and deep learning methods, aiming to improve results with reduced time, costs, effort, and resources. It includes case studies that can be executed on a regular computer and covers theory and application, with a focus on hyperparameter tuning of ML and DL algorithms.

Format: Hardback
Length: 323 pages
Publication date: 02 January 2023
Publisher: Springer Verlag, Singapore

This comprehensive open-access book offers a treasure trove of practical examples showcasing the application of hyperparameter tuning in the real world. It delves deep into the inner workings of machine learning (ML) and deep learning (DL) methods, providing invaluable insights for readers seeking to optimize their performance with minimal time, cost, effort, and resources.

The primary objective of this book is to empower readers with the skills to achieve exceptional results in a remarkably short timeframe, regardless of the resources at their disposal. The case studies presented herein can be executed on a standard desktop or notebook computer, eliminating the need for high-performance computing facilities.

The genesis of this book can be traced back to a study conducted by Bartz & Bartz GmbH for the Federal Statistical Office of Germany (Destatis). Building upon that foundation, the book is tailored to practitioners in industry, as well as researchers, teachers, and students in academia. The content revolves around hyperparameter tuning of ML and DL algorithms, organized into two main sections: theory (Part I) and application (Part II).

Within these sections, essential topics are covered, including:

A comprehensive survey of important model parameters
Detailed analysis of four parameter tuning studies and one extensive global parameter tuning study
Statistical evaluation of the performance of ML and DL methods based on severity
Introduction of a novel consensus-ranking-based approach to aggregate and analyze results from multiple algorithms

The book presents comprehensive analyses of over 30 hyperparameters from six relevant ML and DL methods. Additionally, it includes comprehensive source code, allowing users to reproduce the results and further explore the intricacies of hyperparameter tuning. As such, the book serves as both a handbook and a textbook, offering valuable guidance and knowledge to practitioners and enthusiasts alike.

Weight: 682g
Dimension: 235 x 155 (mm)
ISBN-13: 9789811951695
Edition number: 1st ed. 2023