Chapter 1, A Lap around Automated Machine Learning, presents a detailed overview of AutoML methods by both providing a solid overview for novices and serving as a reference for experienced machine learning practitioners. This chapter starts with the machine learning development life cycle and navigates the problem of hyperparameter optimization that AutoML solves.
Chapter 2, Automated Machine Learning, Algorithms, and Techniques, allows citizen data scientists to build AI solutions without extensive experience. In this chapter, we review the current developments of AutoML in terms of three categories: automated feature engineering (AutoFE), automated model and hyperparameter learning (AutoMHL), and automated deep learning (AutoDL). State-of-the-art techniques adopted in these three categories are presented, including Bayesian optimization, reinforcement learning, evolutionary algorithms, and gradient-based approaches. In this chapter, we'll summarize popular AutoML frameworks and conclude with the current open challenges of AutoML.
Chapter 3, Automated Machine Learning with Open Source Tools and Libraries, teaches you about AutoML open source software (OSS) tools and libraries that automate the entire life cycle of the ideation, conceptualization, development, and deployment of predictive models. From data preparation through model training to validation as well as deployment, these tools do everything with almost zero human intervention. In this chapter, we'll review the major OSS tools, including TPOT, AutoKeras, Auto-Sklearn, Featuretools, H2O AutoML, Auto-PyTorch, Microsoft NNI, and Amazon AutoGluon, and help you to understand the different value propositions and approaches used in each of these libraries.
Chapter 4, Getting Started with Azure Machine Learning, covers Azure Machine Learning, which helps accelerate the end-to-end machine learning life cycle using the power of the Windows Azure platform and services. In this chapter, we will review how to get started with an enterprise-grade machine learning service to build and deploy models that empower developers and data scientists for building, training, and deploying machine learning models faster. With examples, we will set up the groundwork to build and deploy AutoML solutions.
Chapter 5, Automated Machine Learning with Microsoft Azure, reviews in detail and with code examples, how can we automate time-consuming and iterative tasks of model development using an Azure machine learning stack and perform operations such as regression, classification, and time series analysis using Azure AutoML. This chapter will enable you to perform hyperparameter tuning to find the optimal parameters and find the optimal model with Azure AutoML.
Chapter 6, Machine Learning with Amazon Web Services, covers Amazon SageMaker Studio, Amazon SageMaker Autopilot, Amazon SageMaker Ground Truth, and Amazon SageMaker Neo, along with the other AI services and frameworks offered by AWS. As well as hyperscalers (cloud offerings), AWS offers one of the broadest and deepest sets of machine learning services and supporting cloud infrastructure, putting machine learning in the hands of every developer, data scientist, and expert practitioner. AWS offers machine learning services, AI services, deep learning frameworks, and learning tools to build, train, and deploy machine learning models fast.
Chapter 7, Doing Automated Machine Learning with Amazon SageMaker Autopilot, takes us on a deep dive into Amazon SageMaker Studio, using SageMaker Autopilot to run several candidates to figure out the optimal combination of data preprocessing steps, machine learning algorithms, and hyperparameters. The chapter provides a hands-on, illustrative overview of training an inference pipeline, for easy deployment on a real-time endpoint or batch processing.
Chapter 8, Machine Learning with Google Cloud Platform, reviews Google's AI and machine learning offerings. Google Cloud offers innovative machine learning products and services on a trusted and scalable platform. These services include AI Hub, AI building blocks such as sight, language, conversation, and structured data services, and AI Platform. In this chapter, you will become familiar with these offerings and understand how AI Platform supports Kubeflow, Google's open source platform, which lets developers build portable machine learning pipelines with access to cutting-edge Google AI technology such as TensorFlow, TPUs, and TFX tools to deploy your AI applications to production.
Chapter 9, Automated Machine Learning with GCP Cloud AutoML, shows you how to train custom business-specific machine learning models, with minimum effort and machine learning expertise. With hands-on examples and code walk-throughs, we will explore the Google Cloud AutoML platform to create customized deep learning models in natural language, vision, unstructured data, language translation, and video intelligence, without any knowledge of data science or programming.
Chapter 10, AutoML in the Enterprise, presents AutoML in an enterprise setting as a system to automate data science by generating fully automated reports that include an analysis of the data, as well as predictive models and a comparison of their performance. A unique feature of AutoML is that it provides natural-language descriptions of the results, suitable for non-experts in machine learning. We emphasize the operationalization of an MLOps pipeline with a discussion on approaches that perform well on practical problems and determine the best overall approach. The chapter details ideas and concepts behind real-world challenges and provides a journey map to address these problems.
