Agile machine learning: effective machine learning inspired by the agile manifesto

摘要

The application of agile principles to applied machine learning teams is the topic of this book. Machine learning relies heavily on data. This poses new types of challenges, that is, data is generally noisy, follows unknown patterns, and can be biased. The authors tackle these challenges by customizing the agile development principles in order to deliver a high-quality, data-oriented product. The book is comprised of 13 chapters. Each chapter aligns an agile principle with a set of machine learning project requirements. Chapter 1 discusses the need for more data attributes and quality to be ready for early delivery. Chapter 2 proposes new pipelines and models to support the ever-changing requirements in software development. Chapter 3 emphasizes the importance of data verification and continuous deployment to satisfy the continuous delivery requirement.