Designing Machine Learning Systems

Chapter I Unsupervised Machine Learning 1.1 Introducing k-means clustering 6 1.2 Clustering 7 1.3 Kick-starting clustering analysis 8 1.4 Tuning your clustering configurations 12 1.5 Self-organizing maps 17 1. 6 SOM 18 1.7 Employing SOM 19 Chapter II Deep Belief Networks

2.1 Neural networks 25 2.1.1 The composition of a neural network 25 2.2 Network topologies 27 2.2.1 Restricted Boltzmann Machine 31 2.2.2 Introducing the RBM 31 2.2.3 Topology 32 2.2.4 Training 33 2.2.5 Applications of the RBM 34 2.3 Deep belief networks 47 2.3.1 Training a DBN 49 2.3.2 Applying the DBN 50 2.3.3 Validating the DBN 54

Chapter III Stacked Denoising Autoencoders 3.1 Introducing the autoencoder 55 3.2 Topology 57 3.3 Training 58 3.4 Denoising auto encoders 59 3.5 Applying a dA 62 3.6 Stacked Denoising Autoencoders 67 3.7 Applying the SdA 68 3.8 Assessing SdA performance 77

Chapter IV Convolutional Neural Networks

4.1 Introducing the CNN 78 4.2 Understanding the convnet topology 80 4.3 Understanding convolution layers 83 4.4 Understanding pooling layers 87 4.5 Training a convent 87 4.6 Putting it all together 92 4.7 Applying a CNN 97

Chapter V Semi-Supervised Learning

5.1 Introduction 106 5.2 Understanding semi-supervised learning 107 5.3 Semi-supervised algorithms in action 108 5.4 Self-training 109 5.5 Implementing self-training 111 5.6 Improving the selection process 119 5.7 Contrastive Pessimistic Likelihood Estimation 120

Chapter VI Ensemble Methods

6.1 Introducing ensembles 134 6.2 Understanding averaging ensembles 135 6.3 Using bagging algorithms 136 6.4 Using random forests 138 6.5 Applying boosting methods 142 6.6 Using XGBoost 146 6.7 Using stacking ensembles 150 6.8 Understanding model robustness 155 Chapter VII Additional Python Machine Learning Tools

7.1 Alternative development tools 164 7.2 Introduction to Lasagne 164 7.3 Getting to know Lasagne 165 7.4 Introduction to TensorFlow 167 7.5 Getting to know TensorFlow 168 7.6 Using TensorFlow to iteratively improve our models 170 7.7 Knowing when to use these libraries 175