Machine Learning with TensorFlow

by Nishant Shukla

Overview

Summary

Machine Learning with TensorFlow gives readers a solid foundation in machine-learning concepts plus hands-on experience coding TensorFlow with Python.

Purchase of the print book includes a free eBook in PDF, Kindle, and ePub formats from Manning Publications.

About the Technology

TensorFlow, Google's library for large-scale machine learning, simplifies often-complex computations by representing them as graphs and efficiently mapping parts of the graphs to machines in a cluster or to the processors of a single machine.

About the Book

Machine Learning with TensorFlow gives readers a solid foundation in machine-learning concepts plus hands-on experience coding TensorFlow with Python. You'll learn the basics by working with classic prediction, classification, and clustering algorithms. Then, you'll move on to the money chapters: exploration of deep-learning concepts like autoencoders, recurrent neural networks, and reinforcement learning. Digest this book and you will be ready to use TensorFlow for machine-learning and deep-learning applications of your own.

What's Inside

• Matching your tasks to the right machine-learning and deep-learning approaches
• Visualizing algorithms with TensorBoard
• Understanding and using neural networks

About the Reader

Written for developers experienced with Python and algebraic concepts like vectors and matrices.

About the Author

Author Nishant Shukla is a computer vision researcher focused on applying machine-learning techniques in robotics.

Senior technical editor, Kenneth Fricklas, is a seasoned developer, author, and machine-learning practitioner.

Table of Contents

PART 1 - YOUR MACHINE-LEARNING RIG
1. A machine-learning odyssey
2. TensorFlow essentials
PART 2 - CORE LEARNING ALGORITHMS
3. Linear regression and beyond
4. A gentle introduction to classification
5. Automatically clustering data
6. Hidden Markov models
PART 3 - THE NEURAL NETWORK PARADIGM
7. A peek into autoencoders
8. Reinforcement learning
9. Convolutional neural networks
10. Recurrent neural networks
11. Sequence-to-sequence models for chatbots
12. Utility landscape

Product Details

ISBN-13:	9781617293870
Publisher:	Manning
Publication date:	02/12/2018
Edition description:	1st Edition
Pages:	272
Product dimensions:	7.30(w) x 9.20(h) x 0.70(d)

About the Author

Nishant Shukla is a computer vision researcher focused on applying machine-learning techniques in robotics.

Table of Contents

Preface xiii

Acknowledgments xv

About this book xvii

About the author xix

About the cover xx

Part 1 Your Machine-Learning Rig 1

1 A machine-learning odyssey 3

1.1 Machine-learning fundamentals 5

Parameters 7

Learning and inference 8

1.2 Data representation and features 9

1.3 Distance metrics 15

1.4 Types of learning 17

Supervised learning 17

Unsupervised learning 19

Reinforcement learning 19

1.5 TensorFlow 21

1.6 Overview of future chapters 22

1.7 Summary 24

2 TensorFlow essentials 25

2.1 Ensuring that TensorFlow works 27

2.2 Representing tensors 28

2.3 Creating operators 32

2.4 Executing operators with sessions 34

Understanding code as a graph 35

Setting session configurations 36

2.5 Writing code in Jupyter 38

2.6 Using variables 41

2.7 Saving and loading variables 43

2.8 Visualizing data using TensorBoard 44

Implementing a moving average 44

Visualizing the moving average 46

2.9 Summary 49

Part 2 Core Learning Algorithms 51

3 Linear regression and beyond 53

3.1 Formal notation 54

How do you know the regression algorithm is working? 57

3.2 Linear regression 59

3.3 Polynomial model 62

3.4 Regularization 65

3.5 Application of linear regression 69

3.6 Summary 70

4 A gentle introduction to classification 71

4.1 Formal notation 73

4.2 Measuring performance 75

Accuracy 75

Precision and recall 76

Receiver operating characteristic curve 77

4.3 Using linear regression for classification 78

4.4 Using logistic regression 83

Solving one-dimensional logistic regression 84

Solving two-dimensioned logistic regression 87

4.5 Multiclass classifier 90

One-versus-all 91

One-versus-one 92

Softmax regression 92

4.6 Application of classification 96

4.7 Summary 97

5 Automatically clustering data 99

5.1 Traversing files in TensorFlow 100

5.2 Extracting features from audio 102

5.3 K-means clustering 106

5.4 Audio segmentation 109

5.5 Clustering using a sell-organizing map 112

5.6 Application of clustering 117

5.7 Summary 117

6 Hidden Markov models 119

6.1 Example of a not-so-interpretable model 121

6.2 Markov model 121

6.3 Hidden Markov model 124

6.4 Forward algorithm 125

6.5 Viterbi decoding 128

6.6 Uses of hidden Markov models 130

Modeling a video 130

Modeling DNA 130

Madding an image 130

6.7 Application of hidden Markov models 130

6.8 Summary 131

Part 3 Te Neural Network Paradigm 133

7 A peek into autoencoders 135

7.1 Neural networks 136

7.2 Autoencoders 140

7.3 Batch training 145

7.4 Working with images 146

7.5 Application of autoencoders 150

7.6 Summary 151

8 Reinforcement learning 153

8.1 Formal notions 155

Policy 156

Utility 157

8.2 Applying reinforcement learning 158

8.3 Implementing reinforcement learning 160

8.4 Exploring other applications of reinforcement learning 167

8.5 Summary 168

9 Convolutional neural networks 169

9.1 Drawback of neural networks 170

9.2 Convolutional neural networks 171

9.3 Preparing the image 173

Generating filters 176

Convolving using filters 178

Max pooling 181

9.4 Implementing a convolutional neural network in TensorFlow 182

Measuring performance 185

Training the classifier 186

9.5 Tips and tricks to improve performance 187

9.6 Application of convolutional neural networks 188

9.7 Summary 188

10 Recurrent neural networks 189

10.1 Contextual information 190

10.2 Introduction to recurrent neural networks 190

10.3 Implementing a recurrent neural network 192

10.4 A predictive model for time-series data 195

10.5 Application of recurrent neural networks 198

10.6 Summary 199

11 Sequence-to-sequence models for chatbots 201

11.1 Building on classification and RNNs 202

11.2 Seq2seq architecture 205

11.3 Vector representation of symbols 210

11.4 Putting it all together 212

11.5 Gathering dialogue data 220

11.6 Summary 222

12 Utility landscape 223

12.1 Preference model 226

12.2 Image embedding 231

12.3 Ranking images 234

12.4 Summary 239

12.5 What's next? 239

Appendix Installation 241

Index 247