Explainable AI with Python

This book provides a comprehensive presentation of current concepts and techniques to make "machine learning" systems more explainable, with applications to linear and logistic regression, deep learning neural networks, natural language processing, and image recognition. It takes a theoretical and practical perspective, covering model-agnostic methods for XAI and demonstrating how to effectively use ML and XAI in science. The final chapter explains Adversarial Machine Learning and how to do XAI with adversarial examples.

Format: Paperback / softback
Length: 202 pages
Publication date: 29 April 2021
Publisher: Springer Nature Switzerland AG

This comprehensive book offers a thorough exploration of the latest concepts and techniques aimed at enhancing the explainability of "machine learning" systems. By applying these approaches, which encompass linear and logistic regression, deep learning neural networks, natural language processing, and image recognition, among others, these methods can be effectively applied to a wide range of current "machine learning" models.

The field of Machine Learning is witnessing a significant surge in the deployment of artificial agents to carry out crucial tasks that were previously exclusively handled by humans. These agents are being utilized in domains such as healthcare, legal, and finance, among others. While the underlying principles guiding the design of these agents are well-understood, the majority of current deep-learning models remain largely opaque to human comprehension.

Explainable AI with Python aims to bridge this gap in the literature by providing a comprehensive and practical approach to this emerging field. The book takes a theoretical and hands-on approach, equipping readers with the necessary skills to work with tools and code for Explainable AI. It begins by introducing the concept of XAI and highlighting its importance in the field, followed by detailing various approaches to XAI tailored to specific contexts and requirements.

Through practical examples leveraging Python, the book demonstrates how intrinsic interpretable models can be interpreted and how to generate "human-understandable" explanations. Model-agnostic methods for XAI are showcased, enabling the production of explanations without relying on the internal workings of ML models that are often considered "opaque." The authors then extend their exploration to explainable models for Deep Learning, discussing prospective methods for the future.

Taking a practical perspective, the authors demonstrate how to effectively integrate ML and XAI in scientific endeavors. The final chapter delves into Adversarial Machine Learning, exploring how XAI can be performed with adversarial examples.

In conclusion, this book serves as a valuable resource for researchers, practitioners, and students interested in advancing the field of Explainable AI. By offering a comprehensive and practical approach, it empowers readers to work with tools and code for XAI, enabling them to make informed decisions and develop more interpretable and trustworthy "machine learning" systems.

Weight: 336g
Dimension: 155 x 235 x 26 (mm)
ISBN-13: 9783030686390
Edition number: 1st ed. 2021