Hands-On Unsupervised Learning with Python: Implement machine learning and deep learning models using Scikit-Learn, TensorFlow, and more

✍ Scribed by Bonaccorso, Giuseppe

• Publisher: Packt
• Year: 2019
• Tongue: English
• Leaves: 392
• Category: Library

✦ Synopsis

Unsupervised learning is an increasingly important branch of data science, the goal of which is to train models that can learn the structure of a dataset and provide the user with helpful pieces of information about new samples. In many different business sectors (such as marketing, business intelligence, strategy, and so forth), unsupervised learning has always had a primary role in helping the manager to make the best decisions, based both on qualitative and, above all, quantitative approaches. In a world where data is becoming more and more pervasive and storage costs are dropping, the possibility of analyzing real, complex datasets is helping to transform old-fashioned business models into new, more accurate, more responsive, and more effective ones. That's why a data scientist might not have a clear idea about all the possibilities, focusing on the pros and cons of all methods and increasing their knowledge about the best potential strategies for every specific domain. This book is not intended to be an exhaustive resource (which is actually impossible to find), but more of a reference to set you off on your exploration of this world, providing you with different methods that can be immediately employed and evaluated. I hope that readers with different backgrounds will learn worthwhile things for improving their businesses, and that you'll seek more study of this fascinating topic!

✦ Table of Contents

Title Page
Copyright and Credits
Hands-On Unsupervised Learning with Python
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
Download the color images
Conventions used
Get in touch
Reviews
Getting Started with Unsupervised Learning
Technical requirements
Why do we need machine learning?
Descriptive analysis
Diagnostic analysis
Predictive analysis
Prescriptive analysis
Types of machine learning algorithm
Supervised learning algorithms
Supervised hello world!
Unsupervised learning algorithms
Cluster analysis
Generative models
Association rules
Unsupervised hello world!
Semi-supervised learning algorithms
Reinforcement learning algorithms
Why Python for data science and machine learning?
Summary
Questions
Further reading
Clustering Fundamentals
Technical requirements
Introduction to clustering
Distance functions
K-means
K-means++
Analysis of the Breast Cancer Wisconsin dataset
Evaluation metrics
Minimizing the inertia
Silhouette score
Completeness score
Homogeneity score
A trade-off between homogeneity and completeness using the V-measure
Adjusted Mutual Information (AMI) score
Adjusted Rand score
Contingency matrix
K-Nearest Neighbors
Vector Quantization
Summary
Questions
Further reading
Advanced Clustering
Technical requirements
Spectral clustering
Mean shift
DBSCAN
Calinski-Harabasz score
Analysis of the Absenteeism at Work dataset using DBSCAN
Cluster instability as a performance metric
K-medoids
Online clustering
Mini-batch K-means
BIRCH
Comparison between mini-batch K-means and BIRCH
Summary
Questions
Further reading
Hierarchical Clustering in Action
Technical requirements
Cluster hierarchies
Agglomerative clustering
Single and complete linkages
Average linkage
Ward's linkage
Analyzing a dendrogram
Cophenetic correlation as a performance metric
Agglomerative clustering on the Water Treatment Plant dataset
Connectivity constraints
Summary
Questions
Further reading
Soft Clustering and Gaussian Mixture Models
Technical requirements
Soft clustering
Fuzzy c-means
Gaussian mixture
EM algorithm for Gaussian mixtures
Assessing the performance of a Gaussian mixture with AIC and BIC
Component selection using Bayesian Gaussian mixture
Generative Gaussian mixture
Summary
Questions
Further reading
Anomaly Detection
Technical requirements
Probability density functions
Anomalies as outliers or novelties
Structure of the dataset
Histograms
Kernel density estimation (KDE)
Gaussian kernel
Epanechnikov kernel
Exponential kernel
Uniform (or Tophat) kernel
Estimating the density
Anomaly detection
Anomaly detection with the KDD Cup 99 dataset
One-class support vector machines
Anomaly detection with Isolation Forests
Summary
Questions
Further reading
Dimensionality Reduction and Component Analysis
Technical requirements
Principal Component Analysis (PCA)
PCA with Singular Value Decomposition
Whitening
PCA with the MNIST dataset
Kernel PCA
Adding more robustness to heteroscedastic noise with factor analysis
Sparse PCA and dictionary learning
Non-Negative Matrix Factorization
Independent Component Analysis
Topic modeling with Latent Dirichlet Allocation
Summary
Questions
Further reading
Unsupervised Neural Network Models
Technical requirements
Autoencoders
Example of a deep convolutional autoencoder
Denoising autoencoders
Adding noise to the deep convolutional autoencoder
Sparse autoencoders
Adding a sparseness constraint to the deep convolutional autoencoder
Variational autoencoders
Example of a deep convolutional variational autoencoder
Hebbian-based principal component analysis
Sanger's network
An example of Sanger's network
Rubner-Tavan's network
An example of a Rubner-Tavan's network
Unsupervised deep belief networks
Restricted Boltzmann Machines
Deep belief networks
Example of an unsupervised DBN
Summary
Questions
Further reading
Generative Adversarial Networks and SOMs
Technical requirements
Generative adversarial networks
Analyzing a GAN
Mode collapse
Example of a deep convolutional GAN
Wasserstein GANs
Transforming the DCGAN into a WGAN
Self-organizing maps
Example of a Kohonen map
Summary
Questions
Further reading
Assessments
Chapter 1
Chapter 2
Chapter 3
Chapter 4
Chapter 5
Chapter 6
Chapter 7
Chapter 8
Chapter 9
Other Books You May Enjoy
Leave a review - let other readers know what you think