Deep Learning with Microsoft Cognitive Toolkit Quick Start Guide

图书信息

作者	Willem Meints
出版社	Packt Publishing
ISBN	9781789803198
出版时间	2019-03-28
字数	25.4万
分类	进口书,外文原版书,电脑,网络

读书简介

Learn how to train popular deep learning architectures such as autoencoders, convolutional and recurrent neural networks while discovering how you can use deep learning models in your software applications with Microsoft Cognitive Toolkit Key Features * Understand the fundamentals of Microsoft Cognitive Toolkit and set up the development environment * Train different types of neural networks using Cognitive Toolkit and deploy it to production * Evaluate the performance of your models and improve your deep learning skills Book Description Cognitive Toolkit is a very popular and recently open sourced deep learning toolkit by Microsoft. Cognitive Toolkit is used to train fast and effective deep learning models. This book will be a quick introduction to using Cognitive Toolkit and will teach you how to train and validate different types of neural networks, such as convolutional and recurrent neural networks. This book will help you understand the basics of deep learning. You will learn how to use Microsoft Cognitive Toolkit to build deep learning models and discover what makes this framework unique so that you know when to use it. This book will be a quick, no-nonsense introduction to the library and will teach you how to train different types of neural networks, such as convolutional neural networks, recurrent neural networks, autoencoders, and more, using Cognitive Toolkit. Then we will look at two scenarios in which deep learning can be used to enhance human capabilities. The book will also demonstrate how to evaluate your models' performance to ensure it trains and runs smoothly and gives you the most accurate results. Finally, you will get a short overview of how Cognitive Toolkit fits in to a DevOps environment What you will learn * Set up your deep learning environment for the Cognitive Toolkit on Windows and Linux * Pre-process and feed your data into neural networks * Use neural networks to make effcient predictions and recommendations * Train and deploy effcient neural networks such as CNN and RNN * Detect problems in your neural network using TensorBoard * Integrate Cognitive Toolkit with Azure ML Services for effective deep learning Who this book is for Data Scientists, Machine learning developers, AI developers who wish to train and deploy effective deep learning models using Microsoft CNTK will find this book to be useful. Readers need to have experience in Python or similar object-oriented language like C# or Java.

目录

Title Page

Copyright and Credits

Deep Learning with Microsoft Cognitive Toolkit Quick Start Guide

Dedication

About Packt

Why subscribe?

Packt.com

Contributors

About the author

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

Code in Action

Conventions used

Get in touch

Reviews

Getting Started with CNTK

The relationship between AI, machine learning, and deep learning

Limitations of machine learning

How does deep learning work?

The neural network architecture

Artificial neurons

Predicting output with a neural network

Optimizing a neural network

What is CNTK?

Features of CNTK

A high-speed low-level API

Basic building blocks for quickly creating neural networks

Measuring model performance

Loading and processing large datasets

Using models from C# and Java

Installing CNTK

Installing on Windows

Installing Anaconda

Upgrading pip

Installing CNTK

Installing on Linux

Installing Anaconda

Upgrading pip to the latest version

Installing the CNTK package

Using your GPU with CNTK

Enabling GPU usage on Windows

Enabling GPU usage on Linux

Summary

Building Neural Networks with CNTK

Technical requirements

Basic neural network concepts in CNTK

Building neural networks using layer functions

Customizing layer settings

Using learners and trainers to optimize the parameters in a neural network

Loss functions

Model metrics

Building your first neural network

Building the network structure

Choosing an activation function

Choosing an activation function for the output layer

Choosing an activation function for the hidden layers

Picking a loss function

Recording metrics

Training the neural network

Choosing a learner and setting up training

Feeding data into the trainer to optimize the neural network

Checking the performance of the neural network

Making predictions with a neural network

Improving the model

Summary

Getting Data into Your Neural Network

Technical requirements

Training a neural network efficiently with minibatches

Working with small in-memory datasets

Working with numpy arrays

Working with pandas DataFrames

Working with large datasets

Creating a MinibatchSource instance

Creating CTF files

Feeding data into a training session

Taking control over the minibatch loop

Summary

Validating Model Performance

Technical requirements

Choosing a good strategy to validate model performance

Using a hold-out dataset for validation

Using k-fold cross-validation

What about underfitting and overfitting?

Validating performance of a classification model

Using a confusion matrix to validate your classification model

Using the F-measure as an alternative to the confusion matrix

Measuring classification performance in CNTK

Validating performance of a regression model

Measuring the accuracy of your predictions

Measuring regression model performance in CNTK

Measuring performance for out-of-memory datasets

Measuring performance when working with minibatch sources

Measuring performance when working with a manual minibatch loop

Monitoring your model

Using callbacks during training and validation

Using ProgressPrinter

Using TensorBoard

Summary

Working with Images

Technical requirements

Convolutional neural network architecture

Network architecture used for image classification

Working with convolution layers

Working with pooling layers

Other uses for convolutional networks

Building convolutional networks

Building the network structure

Training the network with images

Picking the right combination of layers

Improving model performance with data augmentation

Summary

Working with Time Series Data

Technical requirements

What are recurrent neural networks?

Recurrent neural networks variations

Predicting a single output based on a sequence

Predicting a sequence based on a single sample

Predicting sequences based on sequences

Stacking multiple recurrent layers

How do recurrent neural networks work?

Making predictions with a recurrent neural network

Training a recurrent neural network

Using other recurrent layer types

Working with gated recurrent units

Working with long short-term memory units

When to use other recurrent layer types

Building recurrent neural networks with CNTK

Building the neural network structure

Stacking multiple recurrent layers

Training the neural network with time series data

Predicting output

Summary

Deploying Models to Production

Technical requirements

Using machine learning in a DevOps environment

Keeping track of your data

Training models in a continuous integration pipeline

Deploying models to production

Gathering feedback on your models

Storing your models

Storing model checkpoints to continue training at a later point

Storing portable models for use in other applications

Storing a model in ONNX format

Using ONNX models in C#

Using Azure Machine Learning service to manage models

Deploying Azure Machine Learning service

Exploring the machine learning workspace

Running your first experiment

Deploying your model to production

Summary

Other Books You May Enjoy

Leave a review - let other readers know what you think