Interpretable machine learning in Python
Please sign in so that we can notify you about a reply
The book will help to consciously and effectively work with machine learning models. Introduction to the interpretation of machine learning is given: the importance of the topic, its key concepts and problems are revealed. The methods of interpretation are considered: model-abstracts, anchor and countertactic, for multi-cross forecasting, as well as visualization of spruce neural networks. The issues of configuration for interpretability are disclosed: selection and design of signs, weakening of systematic displacement and causal conclusion, monotonous restrictions, tuning models and resistance to antagonism. The prospects for the development of interpreted machine learning models are shown. Each chapter of the book includes detailed examples of the source code in Python.
The first section of the book is a guide for beginners in interpreting the results of modeling. It gives basic concepts and problems, shows the meaning of machine learning in business. The models of a white box, a black box and a glass box are considered, their comparison was compared and reasonable compromises were proposed.
In the second section, a wide range of interpretation methods is described, also known as methods of explained artificial intelligence, and their use in cases of classification, regression, tabular time series, image or text processing. The modeling results are accompanied by software codes and clear examples.
In the third section, the adjustment of models and work with training data is considered. At the same time, interpretability is ensured by reducing complexity, weakening systematic displacement and increasing reliability. The latest methods of choosing signs, monotonous restrictions, adversarial retraining, etc. are considered.
By the end of this book, you can better understand the models of machine learning and improve them by setting up interpretation
The first section of the book is a guide for beginners in interpreting the results of modeling. It gives basic concepts and problems, shows the meaning of machine learning in business. The models of a white box, a black box and a glass box are considered, their comparison was compared and reasonable compromises were proposed.
In the second section, a wide range of interpretation methods is described, also known as methods of explained artificial intelligence, and their use in cases of classification, regression, tabular time series, image or text processing. The modeling results are accompanied by software codes and clear examples.
In the third section, the adjustment of models and work with training data is considered. At the same time, interpretability is ensured by reducing complexity, weakening systematic displacement and increasing reliability. The latest methods of choosing signs, monotonous restrictions, adversarial retraining, etc. are considered.
By the end of this book, you can better understand the models of machine learning and improve them by setting up interpretation
Author:
Author:Масис Серг
Cover:
Cover:Soft
Category:
- Category:Computer & Technology
Publication language:
Publication Language:Russian
Paper:
Paper:Offset
ISBN:
ISBN:978-5-9775-1735-5
No reviews found