▶What You Will Learn
- Manage AI workflows by using AWS cloud to deploy services that feed smart data products
- Use SageMaker services to create recommendation models
- Scale model training and deployment using Apache Spark on EMR
- Understand how to cluster big data through EMR and seamlessly integrate it with SageMaker
- Build deep learning models on AWS using TensorFlow and deploy them as services
- Enhance your apps by combining Apache Spark and Amazon SageMaker
▶Key Features
- Build machine learning apps on Amazon Web Services (AWS) using SageMaker, Apache Spark and TensorFlow
- Learn model optimization, and understand how to scale your models using simple and secure APIs
- Develop, train, tune and deploy neural network models to accelerate model performance in the cloud
▶Who This Book Is For
This book is for data scientists, machine learning developers, deep learning enthusiasts and AWS users who want to build advanced models and smart applications on the cloud using AWS and its integration services. Some understanding of machine learning concepts, Python programming and AWS will be beneficial.
▶What this book covers
- Chapter 1, Getting Started with Machine Learning for AWS, introduces machine learning to the readers. It explains why it is necessary for data scientists to learn about machine learning and how AWS can help them to solve various real-world problems. We also discuss the AWS services and tools that we will be covered in the book.
- Chapter 2, Classifying Twitter Feeds with Naive Bayes, introduces the basics of the Naive Bayes algorithm and presents a text classification problem that will be addressed by the use of this algorithm and language models. We'll provide examples explaining how to apply Naive Bayes using scikit-learn and Apache Spark on SageMaker's BlazingText. Additionally, we'll explore how to use the ideas behind Bayesian reasoning in more complex scenarios. We will use the Twitter API to stream tweets from two different political candidates and predict who wrote them. We will use scikit-learn, Apache Spark, SageMaker, and BlazingText.
- Chapter 3, Predicting House Value with Regression Algorithms, introduces the basics of regression algorithms and applies them to predict the price of houses given a number of features. We'll also introduce how to use logistic regression for classification problems. Examples in SageMaker for scikit-learn and Apache Spark will be provided. We'll be using the Boston Housing Price dataset https://www.kaggle.com/c/boston-housing/ , along with scikit-learn, Apache Spark, and SageMaker.
- Chapter 4, Predicting User Behavior with Tree-Based Methods, introduces decision trees, random forests, and gradient boosted trees. We will explore how to use these algorithms to predict when users will click on ads. Additionally, we will explain how to use AWS EMR and Apache Spark to engineer models at a large scale. We will use the Adform click prediction dataset (https://doi.org/10.7910/DVN/ TADBY7, Harvard Dataverse, V2). We will use the xgboost, Apache Spark, SageMaker, and EMR libraries.
- Chapter 5, Customer Segmentation Using Clustering Algorithms, introduces the main clustering algorithms by exploring how to apply them for customer segmentation based on consumer patterns. Through AWS SageMaker, we will show how to run these algorithms in skicit-learn and Apache Spark. We will use the e-commerce data from Fabien Daniel (https://www.kaggle.com/fabiendaniel/customer-segmentation/data) and scikit-learn, Apache Spark, and SageMaker.
- Chapter 6, Analyzing Visitor Patterns to Make Recommendations, presents the problem of finding similar users based on their navigation patterns in order to recommend custom marketing strategies. Collaborative filtering and distance-based methods will be introduced with examples in scikit-learn and Apache Spark on AWS SageMaker. We will use Kwan Hui Lim's Theme Park Attraction Visits Dataset (https://sites.google.com/site/limkwanhui/datacode), Apache Spark, and SageMaker.
- Chapter 7, Implementing Deep Learning Algorithms, introduces the reader to the main concepts behind deep learning and explains why it has become so relevant in today's AIpowered products. The aim of this chapter is to not discuss the theoretical details of deep learning, but to explain the algorithms with examples and provide a high-level conceptual understanding of deep learning algorithms. This will give the readers a platform to understand what they are implementing in the next chapters.
- Chapter 8, Implementing Deep Learning with TensorFlow on AWS, goes through a series of practical image-recognition problems and explains how to address them with TensorFlow on AWS. TensorFlow is a very popular deep learning framework that can be used to train deep neural networks. This chapter will explain how TensorFlow can be installed by readers and used to train deep learning models using toy datasets. In this chapter, we'll use the MNIST handwritten digits dataset (http://yann.lecun.com/exdb/mnist/), along with TensorFlow and SageMaker.
- Chapter 9, Image Classification and Detection with SageMaker, revisits the image classification problem we dealt with in the previous chapters, but using SageMaker's image classification algorithm and object detection algorithm. We'll use the following datasets:
‣Caltech256 (http:// www.vision.caltech.edu/Image_Datasets/Caltech256/)
We'll also use AWS Sagemaker.
- Chapter 10, Working with AWS Comprehend, explains the functionality of an AWS tool called Comprehend, which is an NLP tool that performs various useful tasks.
- Chapter 11, Using AWS Rekognition, explains how to use Rekognition, which is an image recognition tool that uses deep learning. The readers will learn an easy way of applying image recognition in their applications.
- Chapter 12, Building Conversational Interfaces Using AWS Lex, explains that AWS Lex is a tool that allows programmers to build conversational interfaces. This chapter introduces the readers to topics such as natural language understanding using deep learning.
- Chapter 13, Creating Clusters on AWS, discusses that one of the key problems in deep learning is understanding how to scale and parallelize learning on multiple machines. In this chapter, we'll examine different ways to create clusters of learners. In particular, we'll focus on how to parallelize deep learning pipelines through distributed TensorFlow and Apache Spark.
- Chapter 14, Optimizing Models in Spark and SageMaker, explains that the models that are trained on AWS can be further optimized to run smoothly in production environments. In this section, we will discuss various tricks that our readers can use to improve the performance of their algorithms.
- Chapter 15, Tuning Clusters for Machine Learning, explains that many data scientists and machine learning practitioners face the problem of scale when attempting to run machine learning data pipelines at scale. In this chapter, we focus primarily on EMR, which is a very powerful tool for running very large machine learning jobs. There are many ways to configure EMR, and not every setup works for every scenario. We will go through the main configurations of EMR and explain how each configuration works for different objectives. Additionally, we'll present other ways to run big data pipelines through AWS.
- Chapter 16, Deploying Models Built on AWS, discusses deployment. At this point, readers will have their models built on AWS and would like to ship them to production. We understand that there are a variety of different contexts in which models should be deployed. In some cases, it's as easy as generating a CSV of actions that would be fed to some system. Often, we just need to deploy a web service that's capable of making predictions. However, there are many times in which we need to deploy these models to complex, low-latency, or edge systems. We will go through the different ways you can deploy machine learning models to production.
