ebook img

Applied Deep Learning with TensorFlow 2: Learn to Implement Advanced Deep Learning Techniques with Python PDF

397 Pages·2022·6.373 MB·English
Save to my drive
Quick download
Download
Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.

Preview Applied Deep Learning with TensorFlow 2: Learn to Implement Advanced Deep Learning Techniques with Python

Applied Deep Learning with TensorFlow 2 Learn to Implement Advanced Deep Learning Techniques with Python — Second Edition — Umberto Michelucci Applied Deep Learning with TensorFlow 2 Learn to Implement Advanced Deep Learning Techniques with Python Second Edition Umberto Michelucci Applied Deep Learning with TensorFlow 2: Learn to Implement Advanced Deep Learning Techniques with Python Umberto Michelucci Dübendorf, Switzerland ISBN-13 (pbk): 978-1-4842-8019-5 ISBN-13 (electronic): 978-1-4842-8020-1 https://doi.org/10.1007/978-1-4842-8020-1 Copyright © 2022 by Umberto Michelucci This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafter developed. Trademarked names, logos, and images may appear in this book. Rather than use a trademark symbol with every occurrence of a trademarked name, logo, or image we use the names, logos, and images only in an editorial fashion and to the benefit of the trademark owner, with no intention of infringement of the trademark. The use in this publication of trade names, trademarks, service marks, and similar terms, even if they are not identified as such, is not to be taken as an expression of opinion as to whether or not they are subject to proprietary rights. While the advice and information in this book are believed to be true and accurate at the date of publication, neither the authors nor the editors nor the publisher can accept any legal responsibility for any errors or omissions that may be made. The publisher makes no warranty, express or implied, with respect to the material contained herein. Managing Director, Apress Media LLC: Welmoed Spahr Acquisitions Editor: Celestin Suresh John Development Editor: James Markham Coordinating Editor: Aditee Mirashi Copy Editor: Kezia Endsley Cover designed by eStudioCalamar Cover image designed by Freepik (www.freepik.com) Distributed to the book trade worldwide by Springer Science+Business Media New York, 1 New York Plaza, Suite 4600, New York, NY 10004-1562, USA. Phone 1-800-SPRINGER, fax (201) 348-4505, e-mail [email protected], or visit www.springeronline.com. Apress Media, LLC is a California LLC and the sole member (owner) is Springer Science + Business Media Finance Inc (SSBM Finance Inc). SSBM Finance Inc is a Delaware corporation. For information on translations, please e-mail [email protected]; for reprint, paperback, or audio rights, please e-mail [email protected]. Apress titles may be purchased in bulk for academic, corporate, or promotional use. eBook versions and licenses are also available for most titles. For more information, reference our Print and eBook Bulk Sales web page at http://www.apress.com/bulk-sales. Any source code or other supplementary material referenced by the author in this book is available to readers on GitHub via the book’s product page, located at www.apress.com/978- 1- 4842- 8019- 5. For more detailed information, please visit http://www.apress.com/source- code. Printed on acid-free paper To my daughter Caterina and my wife Francesca. You are the reason I do what I do. Table of Contents About the Author ���������������������������������������������������������������������������������������������������xiii About the Contributing Author ���������������������������������������������������������������������������������xv About the Technical Reviewer �������������������������������������������������������������������������������xvii Acknowledgments ��������������������������������������������������������������������������������������������������xix Foreword ����������������������������������������������������������������������������������������������������������������xxi Introduction ����������������������������������������������������������������������������������������������������������xxiii Chapter 1: Optimization and Neural Networks ���������������������������������������������������������1 A Basic Understanding of Neural Networks ����������������������������������������������������������������������������������1 The Problem of Learning ���������������������������������������������������������������������������������������������������������������3 A First Definition of Learning ���������������������������������������������������������������������������������������������������3 A Definition of Learning for Neural Networks ��������������������������������������������������������������������������4 Constrained vs� Unconstrained Optimization ���������������������������������������������������������������������������5 Absolute and Local Minima of a Function �������������������������������������������������������������������������������7 Optimization Algorithms ����������������������������������������������������������������������������������������������������������8 Choosing the Right Learning Rate �����������������������������������������������������������������������������������������13 Variations of GD ���������������������������������������������������������������������������������������������������������������������15 How to Choose the Right Mini-Batch Size �����������������������������������������������������������������������������18 [Advanced Section] SGD and Fractals �����������������������������������������������������������������������������������20 Exercises �������������������������������������������������������������������������������������������������������������������������������������21 Conclusion ����������������������������������������������������������������������������������������������������������������������������������25 Chapter 2: Hands-on with a Single Neuron�������������������������������������������������������������27 A Short Overview of a Neuron’s Structure ����������������������������������������������������������������������������������27 A Short Introduction to Matrix Notation ���������������������������������������������������������������������������������30 An Overview of the Most Common Activation Functions �������������������������������������������������������31 v Table of ConTenTs How to Implement a Neuron in Keras �����������������������������������������������������������������������������������������40 Python Implementation Tips: Loops and NumPy �������������������������������������������������������������������41 Linear Regression with a Single Neuron �������������������������������������������������������������������������������������43 The Dataset for the Real-World Example �������������������������������������������������������������������������������43 Linear Regression Model ������������������������������������������������������������������������������������������������������������46 Keras Implementation �����������������������������������������������������������������������������������������������������������47 The Model’s Learning Phase �������������������������������������������������������������������������������������������������49 Model’s Performance Evaluation on Unseen Data �����������������������������������������������������������������51 Logistic Regression with a Single Neuron ����������������������������������������������������������������������������������51 The Dataset for the Classification Problem ���������������������������������������������������������������������������52 Dataset Splitting ��������������������������������������������������������������������������������������������������������������������54 The Logistic Regression Model ����������������������������������������������������������������������������������������������55 The Model’s Learning Phase �������������������������������������������������������������������������������������������������57 The Model’s Performance Evaluation ������������������������������������������������������������������������������������58 Conclusion ����������������������������������������������������������������������������������������������������������������������������������58 Exercises �������������������������������������������������������������������������������������������������������������������������������������59 References ����������������������������������������������������������������������������������������������������������������������������������60 Chapter 3: Feed-Forward Neural Networks ������������������������������������������������������������61 A Short Review of Network’s Architecture and Matrix Notation��������������������������������������������������62 Output of Neurons �����������������������������������������������������������������������������������������������������������������65 A Short Summary of Matrix Dimensions �������������������������������������������������������������������������������66 Hyper-Parameters in Fully Connected Networks �������������������������������������������������������������������67 A Short Review of the Softmax Activation Function for Multiclass Classifications ���������������68 A Brief Digression: Overfitting �����������������������������������������������������������������������������������������������������69 A Practical Example of Overfitting �����������������������������������������������������������������������������������������69 Basic Error Analysis ���������������������������������������������������������������������������������������������������������������76 Implementing a Feed-Forward Neural Network in Keras ������������������������������������������������������������78 Multiclass Classification with Feed-Forward Neural Networks ��������������������������������������������������78 The Zalando Dataset for the Real-World Example �����������������������������������������������������������������79 Modifying Labels for the Softmax Function: One-Hot Encoding ��������������������������������������������83 The Feed-Forward Network Model ����������������������������������������������������������������������������������������85 vi Table of ConTenTs Gradient Descent Variations Performances ���������������������������������������������������������������������������89 Examples of Wrong Predictions ���������������������������������������������������������������������������������������������93 Weight Initialization ���������������������������������������������������������������������������������������������������������������94 Adding Many Layers Efficiently ���������������������������������������������������������������������������������������������97 Comparing Different Networks ��������������������������������������������������������������������������������������������100 Estimating the Memory Requirements of Models ���������������������������������������������������������������������105 General Formula for the Memory Footprint �������������������������������������������������������������������������������107 Exercises �����������������������������������������������������������������������������������������������������������������������������������108 References ��������������������������������������������������������������������������������������������������������������������������������109 Chapter 4: Regularization �������������������������������������������������������������������������������������111 Complex Networks and Overfitting �������������������������������������������������������������������������������������������111 What Is Regularization ��������������������������������������������������������������������������������������������������������������116 About Network Complexity ��������������������������������������������������������������������������������������������������117 ℓ Norm ��������������������������������������������������������������������������������������������������������������������������������118 p ℓ Regularization������������������������������������������������������������������������������������������������������������������118 2 ℓ Regularization������������������������������������������������������������������������������������������������������������������131 1 Are the Weights Really Going to Zero? ��������������������������������������������������������������������������������135 Dropout ��������������������������������������������������������������������������������������������������������������������������������137 Early Stopping ���������������������������������������������������������������������������������������������������������������������141 Additional Methods ��������������������������������������������������������������������������������������������������������������142 Exercises �����������������������������������������������������������������������������������������������������������������������������������143 References ��������������������������������������������������������������������������������������������������������������������������������144 Chapter 5: Advanced Optimizers ��������������������������������������������������������������������������145 Available Optimizers in Keras in TensorFlow 2�5 ����������������������������������������������������������������������145 Advanced Optimizers ����������������������������������������������������������������������������������������������������������������145 Exponentially Weighted Averages ����������������������������������������������������������������������������������������146 Momentum ��������������������������������������������������������������������������������������������������������������������������150 RMSProp ������������������������������������������������������������������������������������������������������������������������������152 Adam �����������������������������������������������������������������������������������������������������������������������������������153 vii Table of ConTenTs Comparison of the Optimizers’ Performance ����������������������������������������������������������������������������154 Small Coding Digression �����������������������������������������������������������������������������������������������������158 Which Optimizer Should You Use? ���������������������������������������������������������������������������������������159 Chapter 6: Hyper-Parameter Tuning ���������������������������������������������������������������������161 Black-Box Optimization �������������������������������������������������������������������������������������������������������������161 Notes on Black-Box Functions ��������������������������������������������������������������������������������������������163 The Problem of Hyper-Parameter Tuning ����������������������������������������������������������������������������������164 Sample Black-Box Problem ������������������������������������������������������������������������������������������������������166 Grid Search ��������������������������������������������������������������������������������������������������������������������������167 Random Search �������������������������������������������������������������������������������������������������������������������172 Coarse to Fine Optimization ������������������������������������������������������������������������������������������������176 Bayesian Optimization ���������������������������������������������������������������������������������������������������������180 Sampling on a Logarithmic Scale ����������������������������������������������������������������������������������������201 Hyper-Parameter Tuning with the Zalando Dataset ������������������������������������������������������������������203 A Quick Note about the Radial Basis Function ��������������������������������������������������������������������������210 Exercises �����������������������������������������������������������������������������������������������������������������������������������211 References ��������������������������������������������������������������������������������������������������������������������������������211 Chapter 7: Convolutional Neural Networks �����������������������������������������������������������213 Kernels and Filters ��������������������������������������������������������������������������������������������������������������������213 Convolution ��������������������������������������������������������������������������������������������������������������������������214 Pooling ���������������������������������������������������������������������������������������������������������������������������������231 Padding �������������������������������������������������������������������������������������������������������������������������������234 Building Blocks of a CNN ����������������������������������������������������������������������������������������������������������235 Convolutional Layers �����������������������������������������������������������������������������������������������������������235 Pooling Layers ���������������������������������������������������������������������������������������������������������������������237 Stacking Layers Together ����������������������������������������������������������������������������������������������������238 An Example of a CNN ����������������������������������������������������������������������������������������������������������������239 Conclusion ��������������������������������������������������������������������������������������������������������������������������������243 Exercises �����������������������������������������������������������������������������������������������������������������������������������243 References ��������������������������������������������������������������������������������������������������������������������������������244 viii Table of ConTenTs Chapter 8: A Brief Introduction to Recurrent Neural Networks ����������������������������245 Introduction to RNNs �����������������������������������������������������������������������������������������������������������������245 Notation �������������������������������������������������������������������������������������������������������������������������������247 The Basic Idea of RNNs �������������������������������������������������������������������������������������������������������248 Why the Name Recurrent ����������������������������������������������������������������������������������������������������249 Learning to Count ����������������������������������������������������������������������������������������������������������������249 Conclusion ��������������������������������������������������������������������������������������������������������������������������������254 Further Readings ����������������������������������������������������������������������������������������������������������������������255 Chapter 9: Autoencoders ��������������������������������������������������������������������������������������257 Introduction �������������������������������������������������������������������������������������������������������������������������������257 Regularization in Autoencoders �������������������������������������������������������������������������������������������260 Feed-Forward Autoencoders �����������������������������������������������������������������������������������������������������260 Activation Function of the Output Layer ������������������������������������������������������������������������������262 The Loss Function ���������������������������������������������������������������������������������������������������������������263 The Reconstruction Error �����������������������������������������������������������������������������������������������������267 Example: Reconstructing Handwritten Digits ����������������������������������������������������������������������268 Autoencoder Applications ���������������������������������������������������������������������������������������������������������270 Dimensionality Reduction ����������������������������������������������������������������������������������������������������270 Classification �����������������������������������������������������������������������������������������������������������������������272 Anomaly Detection ��������������������������������������������������������������������������������������������������������������275 Denoising Autoencoders ������������������������������������������������������������������������������������������������������278 Beyond FFA: Autoencoders with Convolutional Layers �������������������������������������������������������������280 Implementation in Keras �����������������������������������������������������������������������������������������������������������281 Exercises �����������������������������������������������������������������������������������������������������������������������������������283 Further Readings ����������������������������������������������������������������������������������������������������������������������283 Chapter 10: Metric Analysis ���������������������������������������������������������������������������������285 Human-Level Performance and Bayes Error �����������������������������������������������������������������������������286 A Short Story About Human-Level Performance �����������������������������������������������������������������289 Human-Level Performance on MNIST ���������������������������������������������������������������������������������291 Bias �������������������������������������������������������������������������������������������������������������������������������������������291 ix Table of ConTenTs Metric Analysis Diagram �����������������������������������������������������������������������������������������������������������293 Training Set Overfitting �������������������������������������������������������������������������������������������������������������294 Test Set �������������������������������������������������������������������������������������������������������������������������������������295 How to Split Your Dataset ����������������������������������������������������������������������������������������������������������297 Unbalanced Class Distribution: What Can Happen ��������������������������������������������������������������300 Datasets with Different Distributions ����������������������������������������������������������������������������������������306 k-fold Cross Validation ��������������������������������������������������������������������������������������������������������������312 Manual Metric Analysis: An Example ����������������������������������������������������������������������������������������319 Exercises �����������������������������������������������������������������������������������������������������������������������������������329 References ��������������������������������������������������������������������������������������������������������������������������������329 Chapter 11: Generative Adversarial Networks (GANs) ������������������������������������������331 Introduction to GANs �����������������������������������������������������������������������������������������������������������������331 Training Algorithm for GANs ������������������������������������������������������������������������������������������������332 A Practical Example with Keras and MNIST ������������������������������������������������������������������������333 Conditional GANs ����������������������������������������������������������������������������������������������������������������������341 Conclusion ��������������������������������������������������������������������������������������������������������������������������������346 Appendix A: Introduction to Keras ������������������������������������������������������������������������347 S ome History �����������������������������������������������������������������������������������������������������������������������������347 Understanding the Sequential Model ����������������������������������������������������������������������������������������348 Understanding Keras Layers �����������������������������������������������������������������������������������������������������349 Setting the Activation Function �������������������������������������������������������������������������������������������350 Using Functional APIs ����������������������������������������������������������������������������������������������������������������350 Specifying Loss Functions and Metrics ������������������������������������������������������������������������������������352 Putting It All Together and Training �������������������������������������������������������������������������������������������352 Modeling evaluate( ) and predict( ) ��������������������������������������������������������������������������������������������354 Using Callback Functions ����������������������������������������������������������������������������������������������������������354 Saving and Loading Models ������������������������������������������������������������������������������������������������������355 Saving Your Weights Manually ���������������������������������������������������������������������������������������������360 Saving the Entire Model ������������������������������������������������������������������������������������������������������360 Conclusion ��������������������������������������������������������������������������������������������������������������������������������361 x

See more

The list of books you might like

Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.