Machine Learning with Noisy Labels: Definitions, Theory, Techniques and Solutions

Machine Learning and Noisy Labels: Definitions, Theory, Techniques, and Solutions is a comprehensive guide to machine learning with noisy labels, suitable for senior undergraduates, post-graduate students, researchers, and practitioners. It discusses the challenges of using noisy-label training sets and introduces techniques to overcome them, enabling the effective use of limited datasets.

Format: Paperback / softback
Length: 312 pages
Publication date: 01 March 2024
Publisher: Elsevier Science Publishing Co Inc

Machine Learning and Noisy Labels: Definitions, Theory, Techniques, and Solutions is an excellent introduction to machine learning with noisy labels, suitable for senior undergraduates, post-graduate students, researchers, and practitioners utilizing and researching machine learning methods.

Most modern machine learning models based on deep learning techniques rely on carefully curated and cleanly labeled training sets to be reliably trained and deployed. However, the expensive labeling process involved in acquiring such training sets limits the number and size of datasets available for building new models, slowing down progress in the field.

This book defines the different types of label noise, introduces the theory behind the problem, presents the main techniques that enable the effective use of noisy-label training sets, and explains the most accurate methods.

The book begins by defining the different types of label noise, including random noise, systematic noise, and label bias. It then discusses the theory behind the problem, including the concept of label smoothing and the idea of minimizing the cost of label noise.

Next, the book presents the main techniques that enable the effective use of noisy-label training sets, such as label propagation, label smoothing, and active learning. It also discusses the most accurate methods for dealing with label noise, such as deep learning models and ensemble methods.

Throughout the book, real-world examples and case studies are used to illustrate the concepts and techniques discussed. These examples include applications in image recognition, natural language processing, and medical diagnosis.

In conclusion, Machine Learning and Noisy Labels: Definitions, Theory, Techniques, and Solutions is an essential resource for anyone interested in machine learning with noisy labels. It provides a comprehensive and up-to-date introduction to the field, covering the different types of label noise, the theory behind the problem, the main techniques, and the most accurate methods. Whether you are a senior undergraduate, post-graduate student, researcher, or practitioner, this book will help you understand and apply machine learning with noisy labels to solve real-world problems.

Weight: 450g
Dimension: 234 x 156 (mm)
ISBN-13: 9780443154416