Hands-On Artificial Intelligence on Amazon Web Services

By : Tripuraneni, Song

4.2 (6)

Buy this Book

Hands-On Artificial Intelligence on Amazon Web Services

4.2 (6)

By: Tripuraneni, Song

Buy this Book

Overview of this book

From data wrangling through to translating text, you can accomplish this and more with the artificial intelligence and machine learning services available on AWS. With this book, you’ll work through hands-on exercises and learn to use these services to solve real-world problems. You’ll even design, develop, monitor, and maintain machine and deep learning models on AWS. The book starts with an introduction to AI and its applications in different industries, along with an overview of AWS artificial intelligence and machine learning services. You’ll then get to grips with detecting and translating text with Amazon Rekognition and Amazon Translate. The book will assist you in performing speech-to-text with Amazon Transcribe and Amazon Polly. Later, you’ll discover the use of Amazon Comprehend for extracting information from text, and Amazon Lex for building voice chatbots. You will also understand the key capabilities of Amazon SageMaker such as wrangling big data, discovering topics in text collections, and classifying images. Finally, you’ll cover sales forecasting with deep learning and autoregression, before exploring the importance of a feedback loop in machine learning. By the end of this book, you will have the skills you need to implement AI in AWS through hands-on exercises that cover all aspects of the ML model life cycle.

Preface

Who this book is for

What this book covers

To get the most out of this book

Get in touch

Free Chapter

Section 1: Introduction and Anatomy of a Modern AI Application

Introduction to Artificial Intelligence on Amazon Web Services

Technical requirements

What is AI?

Overview of AWS AI offerings

Getting familiar with the AWS CLI

Using Python for AI applications

First project with the AWS SDK

Summary

References

Anatomy of a Modern AI Application

Technical requirements

Understanding the success factors of artificial intelligence applications

Understanding the architecture design principles for AI applications

Understanding the architecture of modern AI applications

Creation of custom AI capabilities

Working with a hands-on AI application architecture

Developing an AI application locally using AWS Chalice

Developing a demo application web user interface

Summary

Further reading

Section 2: Building Applications with AWS AI Services

Detecting and Translating Text with Amazon Rekognition and Translate

Making the world smaller

Understanding the architecture of Pictorial Translator

Setting up the project structure

Implementing services

Implementing RESTful endpoints

Implementing the web user interface

Deploying Pictorial Translator to AWS

Discussing project enhancement ideas

Summary

Further reading

Performing Speech-to-Text and Vice Versa with Amazon Transcribe and Polly

Technical requirements

Technologies from science fiction

Understanding the architecture of Universal Translator

Setting up the project structure

Implementing services

Implementing RESTful endpoints

Implementing the Web User Interface

Deploying the Universal Translator to AWS

Discussing the project enhancement ideas

Summary

References

Extracting Information from Text with Amazon Comprehend

Technical requirements

Working with your Artificial Intelligence coworker

Understanding the Contact Organizer architecture

Setting up the project structure

Implementing services

Implementing RESTful endpoints

Implementing the web user interface

Deploying the Contact Organizer to AWS

Discussing the project enhancement ideas

Summary

Further reading

Building a Voice Chatbot with Amazon Lex

Understanding the friendly human-computer interface

Contact assistant architecture

Understanding the Amazon Lex development paradigm

Setting up the contact assistant bot

Integrating the contact assistant into applications

Summary

Further reading

Section 3: Training Machine Learning Models with Amazon SageMaker

Working with Amazon SageMaker

Technical requirements

Preprocessing big data through Spark EMR

Conducting training in Amazon SageMaker

Deploying the trained Object2Vec and running inference

Running hyperparameter optimization (HPO)

Understanding the SageMaker experimentation service

Bring your own model – SageMaker, MXNet, and Gluon

Bring your own container – R model

Summary

Further reading

Creating Machine Learning Inference Pipelines

Technical requirements

Understanding the architecture of the inference pipeline in SageMaker

Creating features using Amazon Glue and SparkML

Identifying topics by training NTM in SageMaker

Running online versus batch inferences in SageMaker

Summary

Further reading

Discovering Topics in Text Collection

Technical requirements

Reviewing topic modeling techniques

Understanding how the Neural Topic Model works

Training NTM in SageMaker

Deploying the trained NTM model and running the inference

Summary

Further reading

Classifying Images Using Amazon SageMaker

Technical requirements

Walking through convolutional neural and residual networks

Classifying images through transfer learning in Amazon SageMaker

Performing inference through Batch Transform

Summary

Further reading

Sales Forecasting with Deep Learning and Auto Regression

Technical requirements

Understanding traditional time series forecasting

How the DeepAR model works

Understanding model sales through DeepAR

Predicting and evaluating sales

Summary

Further reading

Section 4: Machine Learning Model Monitoring and Governance

Model Accuracy Degradation and Feedback Loops

Technical requirements

Monitoring models for degraded performance

Developing a use case for evolving training data – ad-click conversion

Creating a machine learning feedback loop

Summary

Further reading

What Is Next?

Summarizing the concepts we learned in Part I

Summarizing the concepts we learned in Part II

Summarizing the concepts we learned in Part III

Summarizing the concepts we learned in Part IV

What's next?

Summary

Other Books You May Enjoy

Leave a review - let other readers know what you think

Customer Reviews

4.2 (6)

5 star

66.7%

4 star

16.7%

3 star

2 star

1 star

16.7%

Running hyperparameter optimization (HPO)

It takes data scientists numerous hours and experiments to arrive at an optimal set of hyperparameters that are required for best model performance. This process is mostly based on trial and error.

Although GridSearch is one of the techniques that is traditionally used by data scientists, it suffers from the curse of dimensionality. For example, if we have two hyperparameters, with each taking five possible values, we're looking at calculating objective function 25 times (5 x 5). As the number of hyperparameters grows, the number of times that the objective function is computed blows out of proportion.

Random Search addresses this issue by randomly selecting values of hyperparameters, without doing an exhaustive search of every single combination of hyperparameters. This paper by Bergstra et al. claims that a random search of the...