BUILDING MACHINE LEARNING PROJECTS WITH TENSORFLOW PDFPDF.pdf

发布时间：2022-06-19 发布人：admin 分类：说明书资料大小：13.43M 资料格式：pdf 举报版权申诉

efee5112-8334-40dd-92d1-cff4bafabdd5.pdf-第1页.png

第1页 / 共284页

efee5112-8334-40dd-92d1-cff4bafabdd5.pdf-第2页.png

第2页 / 共284页

efee5112-8334-40dd-92d1-cff4bafabdd5.pdf-第3页.png

第3页 / 共284页

efee5112-8334-40dd-92d1-cff4bafabdd5.pdf-第4页.png

第4页 / 共284页

efee5112-8334-40dd-92d1-cff4bafabdd5.pdf-第5页.png

第5页 / 共284页

efee5112-8334-40dd-92d1-cff4bafabdd5.pdf-第6页.png

第6页 / 共284页

efee5112-8334-40dd-92d1-cff4bafabdd5.pdf-第7页.png

第7页 / 共284页

efee5112-8334-40dd-92d1-cff4bafabdd5.pdf-第8页.png

第8页 / 共284页

Cover

Credits

About the Author

About the Reviewer

www.PacktPub.com

Customer Feedback

Table of Contents

Preface

Chapter 1: Exploring and Transforming Data

TensorFlow's main data structure - tensors

Tensor properties - ranks, shapes, and types

Tensor rank

Tensor shape

Tensor data types

Creating new tensors

From numpy to tensors and vice versa

Getting things done - interacting with TensorFlow

Handling the computing workflow - TensorFlow's data flow graph

Computation graph building

Useful operation object methods

Feeding

Variables

Variable initialization

Saving data flow graphs

Graph serialization language - protocol buffers

Useful methods

Example graph building

Running our programs - Sessions

Basic tensor methods

Simple matrix operations

Reduction

Tensor segmentation

Sequences

Tensor shape transformations

Tensor slicing and joining

Dataflow structure and results visualization - TensorBoard

Command line use

How TensorBoard works

Adding Summary nodes

Common Summary operations

Special Summary functions

Interacting with TensorBoard's GUI

Reading information from disk

Tabulated formats - CSV

The Iris dataset

Reading image data

Loading and processing the images

Reading from the standard TensorFlow format

Summary

Chapter 2: Clustering

Learning from data - unsupervised learning

Clustering

k-means

Mechanics of k-means

Algorithm iteration criterion

k-means algorithm breakdown

Pros and cons of k-means

k-nearest neighbors

Mechanics of k-nearest neighbors

Pros and cons of k-nn

Practical examples for Useful libraries

matplotlib plotting library

Sample synthetic data plotting

scikit-learn dataset module

About the scikit-learn library

Synthetic dataset types

Blobs dataset

Employed method

Circle dataset

Employed method

Moon dataset

Project 1 – k-means clustering on synthetic datasets

Dataset description and loading

Generating the dataset

Model architecture

Loss function description and optimizer loop

Stop condition

Results description

Full source code

k-means on circle synthetic data

Project 2 – nearest neighbor on synthetic datasets

Dataset generation

Model architecture

Loss function description

Stop condition

Results description

Full source code

Summary

Chapter 3: Linear Regression

Univariate linear modelling function

Sample data generation

Determination of the cost function

Least squares

Minimizing the cost function

General minima for least squares

Iterative methods - gradient descent

Example section

Optimizer methods in TensorFlow – the train module

The tf.train.Optimizer class

Other Optimizer instance types

Example 1 – univariate linear regression

Dataset description

Model architecture

Cost function description and Optimizer loop

Stop condition

Results description

Reviewing results with TensorBoard

Full source code

Example – multivariate linear regression

Useful libraries and methods

Pandas library

Dataset description

Model architecture

Loss function description and Optimizer loop

Stop condition

Results description

Full source code

Summary

Chapter 4: Logistic Regression

Problem description

Logistic function predecessor – the logit functions

Bernoulli distribution

Link function

Logit function

The importance of the logit inverse

The logistic function

Logistic function as a linear modeling generalization

Final estimated regression equation

Properties of the logistic function

Loss function

Multiclass application – softmax regression

Cost function

Data normalization for iterative methods

One hot representation of outputs

Example 1 – univariate logistic regression

Useful libraries and methods

TensorFlow's softmax implementation

Dataset description and loading

The CHDAGE dataset

CHDAGE dataset format

Dataset loading and preprocessing implementation

Model architecture

Loss function description and optimizer loop

Stop condition

Results description

Fitting function representations across epochs

Full source code

Graphical representation

Example 2 – Univariate logistic regression with skflow

Useful libraries and methods

Dataset description

Model architecture

Results description

Full source code

Summary

Chapter 5: Simple FeedForward Neural Networks

Preliminary concepts

Artificial neurons

Original example – the Perceptron

Perceptron algorithm

Neural network layers

Neural Network activation functions

Gradients and the back propagation algorithm

Minimizing loss function: Gradient descent

Neural networks problem choice - Classification vs Regression

Useful libraries and methods

TensorFlow activation functions

TensorFlow loss optimization methods

Sklearn preprocessing utilities

First project – Non linear synthetic function regression

Dataset description and loading

Dataset preprocessing

Modeling architecture - Loss Function description

Loss function optimizer

Accuracy and Convergence test

Example code

Results description

Second project – Modeling cars fuel efficiency with non linear regression

Dataset description and loading

Dataset preprocessing

Modeling architecture

Convergency test

Results description

Third project – Learning to classify wines: Multiclass classification

Dataset description and loading

Dataset preprocessing

Modeling architecture

Loss function description

Loss function optimizer

Convergence test

Results description

Full source code

Summary

Chapter 6: Convolutional Neural Networks

Origin of convolutional neural networks

Getting started with convolution

Continuous convolution

Discrete convolution

Kernels and convolutions

Interpretation of the convolution operations

Applying convolution in TensorFlow

Other convolutional operations

Sample code – applying convolution to a grayscale image

Sample kernels results

Subsampling operation – pooling

Properties of subsampling layers

Invariance property

Subsampling layers implementation performance.

Applying pool operations in TensorFlow

Other pool operations

Sample code

Improving efficiency – dropout operation

Applying the dropout operation in TensorFlow

Sample code

Convolutional type layer building methods

Convolutional layer

Subsampling layer

Example 1 – MNIST digit classification

Dataset description and loading

Dataset preprocessing

Modelling architecture

Loss function description

Loss function optimizer

Accuracy test

Result description

Full source code

Example 2 – image classification with the CIFAR10 dataset

Dataset description and loading

Dataset preprocessing

Modelling architecture

Loss function description and optimizer

Training and accuracy tests

Results description

Full source code

Summary

Chapter 7: Recurrent Neural Networks and LSTM

Recurrent neural networks

Exploding and vanishing gradients

LSTM neural networks

The gate operation – a fundamental component

Operation steps

Part 1 – set values to forget (input gate)

Part 2 – set values to keep, change state

Part 3 – output filtered cell state

Other RNN architectures

TensorFlow LSTM useful classes and methods

class tf.nn.rnn_cell.BasicLSTMCell

class MultiRNNCell(RNNCell)

learn.ops.split_squeeze(dim, num_split, tensor_in)

Example 1 – univariate time series prediction with energy consumption data

Dataset description and loading

Dataset preprocessing

Modelling architecture

Loss function description

Convergency test

Results description

Full source code

Example 2 – writing music “a la” Bach

Character level models

Character sequences and probability representation

Encoding music as characters – the ABC music format

ABC format data organization

Useful libraries and methods

Saving and restoring variables and models

Loading and saving pseudocode

Variable saving

Variable restoring

Dataset description and loading

Network Training

Dataset preprocessing

Vocabulary definition

Modelling architecture

Loss function description

Stop condition

Results description

Full source code

Summary

Chapter 8: Deep Neural Networks

Deep neural network definition

Deep network architectures through time

LeNet 5

Alexnet

Main features

The original inception model

GoogLenet (Inception V1)

Batch normalized inception (V2)

Inception v3

Residual Networks (ResNet)

Other deep neural network architectures

Example – painting with style – VGG style transfer

Useful libraries and methods

Dataset description and loading

Dataset preprocessing

Modeling architecture

Loss functions

Content loss function

Style loss function

Loss optimization loop

Convergency test

Program execution

Full source code

Summary

Chapter 9: Running Models at Scale – GPU and Serving

GPU support on TensorFlow

Log device placement and device capabilities

Querying the computing capabilities

Selecting a CPU for computing

Device naming

Example 1 – assigning an operation to the GPU

Example 2 – calculating Pi number in parallel

Solution implementation

Source code

Distributed TensorFlow

Technology components

Jobs

Tasks

Servers

Combined overview

Creating a TensorFlow cluster

ClusterSpec definition format

Creating tf.Train.Server

Cluster operation – sending computing methods to tasks

Sample distributed code structure

Example 3 – distributed Pi calculation

Server script

Client script

Full source code

Example 4 – running a distributed model in a cluster

Sample code

Summary

Chapter 10: Library Installation and Additional Tips

Linux installation

Initial requirements

Ubuntu preparation tasks (need to apply before any method)

Pip Linux installation method

CPU version

Testing your installation

GPU support

Virtualenv installation method

Environment test

Docker installation method

Installing Docker

Allowing Docker to run with a normal user

Reboot

Testing the Docker installation

Run the TensorFlow container

Linux installation from source

Installing the Git source code version manager

Git installation in Linux (Ubuntu 16.04)

Installing the Bazel build tool

Adding the Bazel distribution URI as a package source

Updating and installing Bazel

Installing GPU support (optional)

Installing CUDA system packages

Creating alternative locations

Installing cuDNN

Clone TensorFlow source

Configuring TensorFlow build

Building TensorFlow

Testing the installation

Windows installation

Classic Docker toolbox method

Installation steps

Downloading the Docker toolbox installer

Creating the Docker machine

MacOS X installation

Install pip

Summary

Index

Building Machine Learning Projects with TensorFlow Engaging projects that will teach you how complex data can be exploited to gain the most insight Rodolfo Bonnin BIRMINGHAM - MUMBAI

Building Machine Learning Projects with TensorFlow Copyright © 2016 Packt Publishing All rights reserved. No part of this book may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, without the prior written permission of the publisher, except in the case of brief quotations embedded in critical articles or reviews. Every effort has been made in the preparation of this book to ensure the accuracy of the information presented. However, the information contained in this book is sold without warranty, either express or implied. Neither the author, nor Packt Publishing, and its dealers and distributors will be held liable for any damages caused or alleged to be caused directly or indirectly by this book. Packt Publishing has endeavored to provide trademark information about all of the companies and products mentioned in this book by the appropriate use of capitals. However, Packt Publishing cannot guarantee the accuracy of this information. First published: November 2016 Production reference: 2220317 Published by Packt Publishing Ltd. Livery Place 35 Livery Street Birmingham B3 2PB, UK. ISBN 978-1-78646-658-7 www.packtpub.com

Credits Author Rodolfo Bonnin Reviewer Niko Gamulin Copy Editor Safis Editing Project Coordinator Nidhi Joshi Commissioning Editor Veena Pagare Proofreader Safis Editing Acquisition Editor Namrata Patil Indexer Mariammal Chettiyar Content Development Editor Siddhesh Salvi Graphics Disha Haria Technical Editor Danish Shaikh Dharmendra Yadav Production Coordinator Arvindkumar Gupta

About the Author Rodolfo Bonnin is a systems engineer and PhD student at Universidad Tecnológica Nacional, Argentina. He also pursued parallel programming and image understanding postgraduate courses at Uni Stuttgart, Germany. He has done research on high performance computing since 2005 and began studying and implementing convolutional neural networks in 2008,writing a CPU and GPU - supporting neural network feed forward stage. More recently he's been working in the field of fraud pattern detection with Neural Networks, and is currently working on signal classification using ML techniques. To my wife and kids and the patience they demonstrated during the writing of this book. Also to the reviewers, who helped give professionalism to this work, and Marcos Boaglio for facilitating equipment to cover the installation chapter. Ad Maiorem Dei Gloriam.

About the Reviewer Niko Gamulin is a senior software engineer at CloudMondo, a US-based startup, where he develops and implements predictive behavior models for humans and systems. Previously he has developed deep learning models to solve various challenges. He received his PhD in Electrical Engineering from University of Ljubljana in 2015. His research focused on creation of machine learning models for churn prediction. I would like to thank my wonderful daughter Agata, who inspires me to gain more understanding about the learning process and Ana for being the best wife in the world.

www.PacktPub.com For support files and downloads related to your book, please visit www.PacktPub.com. Did you know that Packt offers eBook versions of every book published, with PDF and ePub files available? You can upgrade to the eBook version at www.PacktPub.com and as a print book customer, you are entitled to a discount on the eBook copy. Get in touch with us at service@packtpub.com for more details. At www.PacktPub.com, you can also read a collection of free technical articles, sign up for a range of free newsletters and receive exclusive discounts and offers on Packt books and eBooks. h t t p s ://w w w . p a c k t p u b . c o m /m a p t Get the most in-demand software skills with Mapt. Mapt gives you full access to all Packt books and video courses, as well as industry-leading tools to help you plan your personal development and advance your career. Why subscribe? Fully searchable across every book published by Packt Copy and paste, print, and bookmark content On demand and accessible via a web browser

Customer Feedback Thanks for purchasing this Packt book. At Packt, quality is at the heart of our editorial process. To help us improve, please leave us an honest review on this book's Amazon page at https://www.amazon.com/dp/1786466589/. If you'd like to join our team of regular reviewers, you can e-mail us at customerreviews@packtpub.com. We award our regular reviewers with free eBooks and videos in exchange for their valuable feedback. Help us be relentless in improving our products!

分享到：

赞收藏

资料库

BUILDING MACHINE LEARNING PROJECTS WITH TENSORFLOW PDFPDF.pdf

相关推荐

人工智能

热门标签

最新资料