Mastering Machine Learning on AWS电子书

Dedication

About Packt

Why subscribe?

Packt.com

Contributors

About the authors

About the reviewer

Packt is searching for authors like you

Preface

Who this book is for

What this book covers

To get the most out of this book

Download the example code files

Download the color images

Conventions used

Get in touch

Reviews

Section 1: Machine Learning on AWS

Getting Started with Machine Learning for AWS

How AWS empowers data scientists

Using AWS tools for ML

Identifying candidate problems that can be solved using ML

The ML project life cycle

Data gathering

Evaluation metrics

Algorithm selection

Deploying models

Summary

Exercises

Section 2: Implementing Machine Learning Algorithms at Scale on AWS

Classifying Twitter Feeds with Naive Bayes

Classification algorithms

Feature types

Nominal features

Ordinal features

Continuous features

Naive Bayes classifier

Bayes' theorem

Posterior

Likelihood

Prior probability

Evidence

How the Naive Bayes algorithm works

Classifying text with language models

Collecting the tweets

Preparing the data

Building a Naive Bayes model through SageMaker notebooks

Naïve Bayes model on SageMaker notebooks using Apache Spark

Using SageMaker's BlazingText built-in ML service

Naive Bayes – pros and cons

Summary

Exercises

Predicting House Value with Regression Algorithms

Predicting the price of houses

Understanding linear regression

Linear least squares estimation

Maximum likelihood estimation

Gradient descent

Evaluating regression models

Mean absolute error

Mean squared error

Root mean squared error

R-squared

Implementing linear regression through scikit-learn

Implementing linear regression through Apache Spark

Implementing linear regression through SageMaker's Linear Learner

Understanding logistic regression

Logistic regression in Spark

Pros and cons of linear models

Summary

Predicting User Behavior with Tree-Based Methods

Understanding decision trees

Recursive splitting

Types of decision trees

Cost functions

Gini Impurity

Information gain

Criteria to stop splitting trees

Understanding random forest algorithms

Understanding gradient-boosting algorithms

Predicting clicks on log streams

Introduction to Elastic MapReduce (EMR)

Training with Apache Spark on EMR

Getting the data

Preparing the data

Categorical encoding

One-hot encoding

Training a model

Evaluating our model

Area under the ROC curve

Area under the precision-recall curve

Training tree ensembles on EMR

Training gradient-boosted trees with the SageMaker services

Preparing the data

Training with SageMaker XGBoost

Applying and evaluating the model

Summary

Exercises

Customer Segmentation Using Clustering Algorithms

Understanding how clustering algorithms work

k-means clustering

Euclidean distance

Manhattan distance

Hierarchical clustering

Agglomerative clustering

Divisive clustering

Clustering with Apache Spark on EMR

Clustering with Spark and SageMaker on EMR

Understanding the purpose of the IAM role

Summary

Exercises

Analyzing Visitor Patterns to Make Recommendations

Making theme park attraction recommendations through Flickr data

Collaborative filtering

Memory-based approach

Model-based approach

Matrix factorization

Stochastic gradient descent

Alternating least squares

Finding recommendations through Apache Spark's ALS

Data gathering and exploration

Training the model

Getting recommendations

Recommending attractions through SageMaker FMs

Preparing the dataset for learning

Training the model

Getting recommendations

Summary

Exercises

Section 3: Deep Learning

Implementing Deep Learning Algorithms

Understanding deep learning

Applications of deep learning

Self-driving cars

Learning to play video games using a deep learning algorithm

Understanding deep learning algorithms

Neural network algorithms

Activation functions

Backpropagation

Introduction to deep neural networks

Understanding convolutional neural networks

Summary

Exercises

Implementing Deep Learning with TensorFlow on AWS

Introducing TensorFlow

TensorFlow as a general machine learning library

Training and serving the TensorFlow model through SageMaker

Creating a custom neural net with TensorFlow

Summary

Exercises

Image Classification and Detection with SageMaker

Introducing Amazon SageMaker for image classification

Training a deep learning model using Amazon SageMaker

Classifying images using Amazon SageMaker

Summary

Exercises

Section 4: Integrating Ready-Made AWS Machine Learning Services

Working with AWS Comprehend

Introducing Amazon Comprehend

Accessing Amazon Comprehend

Named-entity recognition using Comprehend

Sentiment analysis using Comprehend

Text classification using Comprehend

Summary

Exercises

Using AWS Rekognition

Introducing Amazon Rekognition

Implementing object and scene detection

Implementing facial analysis

Other Rekognition services

Image moderation

Celebrity recognition

Face comparison

Summary

Exercises

Building Conversational Interfaces Using AWS Lex

Introducing Amazon Lex

Building a custom chatbot using Amazon Lex

Summary

Exercises

Section 5: Optimizing and Deploying Models through AWS

Creating Clusters on AWS

Choosing your instance types

On-demand versus spot instance pricing

Reserved pricing

Amazon Machine Images (AMIs)

Deep learning hardware

Distributed deep learning

Model parallelization versus data parallelization

Distributed TensorFlow

Distributed learning through Apache Spark

Data parallelization

Model parallelization

Distributed hyperparameter tuning

Distributed predictions at scale

Parallelization in SageMaker

Summary

Optimizing Models in Spark and SageMaker

The importance of model optimization

Automatic hyperparameter tuning

Hyperparameter tuning in Apache Spark

Hyperparameter tuning in SageMaker

Summary

Exercises

Tuning Clusters for Machine Learning

Introduction to the EMR architecture

Apache Hadoop

Apache Spark

Apache Hive

Presto

Apache HBase

Yet Another Resource Negotiator (YARN)

Tuning EMR for different applications

Configuring application properties

Maximize Resource Allocation

The AWS Glue Catalog

Managing data pipelines with Glue

Creating tables with Glue

Accessing Glue tables in Spark

Summary

Deploying Models Built in AWS

SageMaker model deployment

Apache Spark model deployment

Summary

Exercises

Appendix: Getting Started with AWS

Other Books You May Enjoy

Leave a review - let other readers know what you think