Search

machineLearning­OUP­SRIDHAR­2021­INTRO.pdf
•0 likes•486 views

D devanthanv2008 Follow

This document provides an introduction to a textbook on machine learning. It is authored by Dr. S.

1of35

Download now

More Related Content

machineLearning­OUP­SRIDHAR­2021­INTRO.pdf
1. 1 Dr S. Sridhar Professor, Department of Information Science and Technology, College of Engineering, Guindy
Campus, Anna University, Chennai Dr M. Vijayalakshmi Associate Professor, Department of Information Science and
Technology, College of Engineering, Guindy Campus, Anna University, Chennai Learning MACHINE ML_FM.indd 1 27­
Mar­21 8:45:05 AM © Oxford University Press. All rights reserved. O x f o r d U n i v e r s i t y P r e s s
2. 3 Oxford University Press is a department of the University of Oxford. It furthers the University’s objective of
excellence in research, scholarship, and education by publishing worldwide. Oxford is a registered trade mark of
Oxford University Press in the UK and in certain other countries. Published in India by Oxford University Press 22
Workspace, 2nd Floor, 1/22 Asaf Ali Road, New Delhi 110002 © Oxford University Press 2021 The moral rights of the
author/s have been asserted. First Edition published in 2021 All rights reserved. No part of this publication may be
reproduced, stored in a retrieval system, or transmitted, in any form or by any means, without the prior permission in
writing of Oxford University Press, or as expressly permitted by law, by licence, or under terms agreed with the
appropriate reprographics rights organization. Enquiries concerning reproduction outside the scope of the above
should be sent to the Rights Department, Oxford University Press, at the address above. You must not circulate this
work in any other form and you must impose this same condition on any acquirer. ISBN­13 (print edition): 978­0­19­
012727­5 ISBN­10 (print edition): 0­19­012727­9 Typeset in Palatino Linotype by B2K­Bytes 2 Knowledge, Tamil Nadu
Printed in India by Cover image: © whiteMocca/Shutterstock For product information and current price, please visit
www.india.oup.com Third­party website addresses mentioned in this book are provided by Oxford University Press in
good faith and for information only. Oxford University Press disclaims any responsibility for the material contained
therein. ML_FM.indd 2 27­Mar­21 8:45:05 AM © Oxford University Press. All rights reserved. O x f o r d U n i v e r s i t y P r
ess
3. Dedicated to my mother, late Mrs Parameswari Sundaramurthy, and my mother­in­law, late Mrs Renuga Nagarajan,
whose encouragement and moral support motivated me to start writing this book when they were alive and
remained an eternal motivation even a몭er their demise in completing this book. – Dr S. Sridhar Dedicated with love
and a몭ection to my family, Husband (Mathan), Sons (Sam and Ayden), and Mom (Vennila). – Dr M. Vijayalakshmi
ML_FM.indd 3 27­Mar­21 8:45:05 AM © Oxford University Press. All rights reserved. O x f o r d U n i v e r s i t y P r e s s
4. Preface Can machines learn like human beings? This question has been posed over many decades and the search
for an answer resulted in the domain of artificial intelligence. Informally, learning is nothing but adaptability to the
environment. Human beings can adapt to the environment using the learning process. Machine Learning is a branch
of artificial intelligence which is defined as “the field of study that gives computers the ability to learn without any
explicit programming”. Unlike other user­defined programs, machine learning programs try to learn from data
automatically. Initially, computer scientists and researchers used automatic learning through logical reasoning.
However, much progress could not be made using logic. Eventually, machine learning became popular due to the
success of data­driven learning algorithms. Human beings have always considered themselves as the superior
species in object recog­ nition. While machines can crunch numbers in seconds, human beings have shown
superiority in recognizing objects. However, recent applications in deep learning show that computers are also good
in facial recognition. The recent developments in machine learning such as object recognition, social media analysis
like sentiment analysis, recommendation systems including Amazon’s book recommendation, innovations, for
example driverless cars, and voice assistance systems, which include Amazon’s Alexa, Microso몭’s Cortana and Google
Assistant, have created more awareness about machine learning. The availability of smart phones, IoT, and cloud
technology have brought these machine learning technologies to daily life. Business and government organizations
benefit from machine learning technology. These organizations traditionally have a huge amount of data. Social
networks such as Twitter, YouTube, and Facebook generate data in terms of Terabytes (TB), Exabytes (EB), and
Zettabytes (ZB). In technologies like IoT, sensors generate a huge amount of data independent of any human
intervention. A sudden interest has been seen in using machine learning algorithms to extract knowledge from these
data automatically. Why? The reason being that extracted knowledge can be useful for prediction and helps in better
decision making. This facilitates the development of many knowledge­based and intelligent applications. Therefore,
awareness of basic machine learning is a must for students and researchers, computer scientists and professionals,
data analysts and data scientists. Historically, these organizations used statistics to analyze these data, but statistics
could not be applied on big data. The need to process enormous amount of data poses a challenge as new
techniques are required to process this voluminous data, and hence, machine learning is the driving force for many
fields such as data science, data analysis, data analytics, data mining, and big data analytics. Scope of the Book Our
aim has been to provide a first level textbook that follows a simple algorithmic approach and comes with numerical
problems to explain the concepts of machine learning. This book stems from the experience of the authors in
teaching the students at the Anna University and the National Institute of Technology for over three decades. It
targets the undergraduate and post­graduate students in computer science, information technology, and in general,
engineering students. This book is also useful for the ones who study data science, data analysis, data analytics, and
data mining. ML_FM.indd 4 27­Mar­21 8:45:05 AM © Oxford University Press. All rights reserved. O x f o r d U n i v e r s i
tyPress
5. Preface v This book comprises chapters covering the concepts of machine learning, a laboratory manual that
consists of 25 experiments implemented in Python, and appendices on the basics of Python and Python packages.
The theory part includes many numerical problems, review questions and pedagogical support such as crossword
and word search. Additional material is available as online resources for better understanding of the subject. Key
Features of the Book • Uses only minimal mathematics to understand the machine learning algorithms covered in
the book • Follows an algorithmic approach to explain the basics of machine learning • Comes with various
numerical problems to emphasize on the important concepts of data analytics • Includes a laboratory manual for
implementing machine learning concepts in Python environment • Has two appendices covering the basics of
Python and Python packages • Focuses on pedagogy like chapter­end review and numerical questions, crosswords
and jumbled word searches • Illustrates important and latest concepts like deep learning, regression analysis,
support vector machines, clustering algorithms, and ensemble learning Content and Organization The book is
divided into 16 chapters and three appendices. The appendices A, B, and C can be accessed through the QR codes
provided in the table of contents. Chapter 1 introduces the Basic Concepts of Machine Learning and explores its
relationships with other domains. This chapter also explores the machine learning types and applications. Chapter 2
of this book is about Understanding Data, which is crucial for data­driven machine learning algorithms. The
mathematics that is necessary for understanding data such as linear algebra and statistics covering univariate,
bivariate and multivariate statistics are introduced in this chapter. This chapter also includes feature engineering and
dimensionality reduction techniques. Chapter 3 covers the Basic Concepts of Learning. This chapter discusses about
theoretical aspects of learning such as concept learning, version spaces, hypothesis, and hypothesis space. It also
introduces learning frameworks like PAC learning, mistake bound model, and VC dimensions. Chapter 4 is about
Similarity Learning. It discusses instance­based learning, nearest­neighbor learning, weighted k­nearest algorithms,
nearest centroid classifier, and locally weighted regression (LWR) algorithms. Chapter 5 introduces the basics of
Regression. The concepts of linear regression and non­linear regression are discussed in this chapter. It also covers
logistic regression. Finally, this chapter outlines the recent algorithms like Ridge, Lasso, and Elastic Net regression.
Chapter 6 throws light on the concept of Decision Tree Learning. The concept of information theory, entropy, and
information gain are discussed in this chapter. The basics of tree construction ML_FM.indd 5 27­Mar­21 8:45:05 AM ©
Oxford University Press. All rights reserved. O x f o r d U n i v e r s i t y P r e s s
6. vi Preface algorithms like ID3, C4.5, CART, and Regression Trees and its illustration are included in this chapter.
The decision tree evaluation is also introduced here. Chapter 7 discusses Rule­based Learning. This chapter
illustrates rule generation. The sequential covering algorithms like PRISM and FOIL are introduced here. This chapter
also discusses analytical learning, explanation­based learning, and active learning mechanisms. An outline of
association rule mining is also provided in this chapter. Chapter 8 introduces the basics of Bayesian model. The
chapter covers the concepts of classi­ fication using the Bayesian principle. Naïve Bayesian classifier and Continuous
Features classifi­ cation are introduced in this chapter. The variants of Bayesian classifier are also discussed. Chapter
9 discusses Probabilistic Graphical Models. The discussion of the Bayesian Belief network construction and its
inference mechanism are included in this chapter. Markov chain and Hidden Markov Model (HMM) are also
introduced along with the associated algorithms. Chapter 10 introduces the basics of Artificial Neural Networks
(ANN). The chapter introduces the concepts of neural networks such as neurons, activation functions, and ANN types.
Perceptron, back­propagation neural networks, Radial Basis Function Neural Network (RBFNN), Self­ Organizing
Feature Map (SOFM) are covered here. The chapter ends with challenges and some applications of ANN. Chapter 11
covers Support Vector Machines (SVM). This chapter begins with a gentle intro­ duction of linear discriminant analysis
and then covers the concepts of SVM such as margins, kernels, and its associated optimization theory. The hard
margin and so몭 margin SVMs are intro­ duced here. Finally, this chapter ends with support vector regression. Chapter
12 introduces Ensemble Learning. It covers meta­classifiers, the concept of voting, bootstrap resampling, bagging,
and random forest and stacking algorithms. This chapter ends with the popular AdaBoost algorithm. Chapter 13
discusses Cluster Analysis. Hierarchical clustering algorithms and partitional clustering algorithms like k­means
algorithm are introduced in this chapter. In addition, the outline of density­based, grid­based and probability­based
approaches like fuzzy clustering and EM algorithm is provided. This chapter ends with an evaluation of clustering
algorithms. Chapter 14 covers the concept of Reinforcement Learning. This chapter starts with the idea of
reinforcement learning, multi­arm bandit problem and Markov Decision Process (MDP). It then introduces model­
based (passive learning) and model­free methods. The Q­Learning and SARSA concepts are also covered in this
chapter. Chapter 15 is about Genetic Algorithms. The concepts of genetic algorithms and genetic algo­ rithm
components along with simple examples are present in this chapter. Evolutionary compu­ tation, like simulated
annealing, and genetic programming are outlined at the end of this chapter. Chapter 16 discusses Deep Learning.
CNN and RNN are explained in this chapter. Long Short­Term Memory (LSTM) and Gated Recurrent Unit (GRU) are
outlined here. Additional web contents are provided for a thorough understanding of deep learning. Appendix A
discusses Python basics. Appendix B covers Python packages that are necessary to implement the machine learning
algorithms. The packages like Numpy, Pandas, Matplotlib, Scikit­learn and Keras are outlined in this appendix.
Appendix C o몭ers 25 laboratory experiments covering the concepts of the textbook. ML_FM.indd 6 27­Mar­21 8:45:05
AM © Oxford University Press. All rights reserved. O x f o r d U n i v e r s i t y P r e s s
7. Acknowledgments The Lord is my strength and my shield; my heart trusted in him, and I am helped: therefore my
heart greatly rejoiceth; and with my song will I praise him. [Psalm 28:7] First and foremost, the authors express the
gratitude with praises and thanks to Lord God who made this thing happen. This book would not have been possible
without the help of many friends, students, colleagues, and well­wishers. The authors thank all the students who
motivated them by asking them the right questions at appropriate time. The authors also thank the colleagues of the
Department of Information Science and Technology, Anna University, SRM University, and National Institute of
Technology for supporting them with their constant reviews and sugges­ tions. The authors acknowledge the
support of the reviewers for providing critical suggestions to improve the quality of the book. We thank the editorial
team of the Oxford University Press, India for their support in bringing the book to realization. The authors thank the
members of the marketing team too for their continuous support and encouragement during the development of
this book. Dr Sridhar thanks his wife, Dr Vasanthy, and daughters, Dr Shobika and Shree Varshika, for being patient
and accommodative. He also wants to recollect the contributions made by his mother, Mrs Parameswari
Sundaramurthy, and mother­in­law, Mrs Renuga Nagarajan, who provided constant encouragement and remained
an eternal motivation in completing the book. Dr Vijayalakshmi would like to thank her husband, Mr Mathan, Sons,
Sam and Ayden, and her mother, Ms Vennila, for supporting her in completing this book successfully. Any comments
and suggestions for further improvement of the book are welcome; please send them at [email protected]
and his website www.drssridhar.com, or [email protected]. S. Sridhar M. Vijayalakshmi The Oxford University Press,
India would like to thank all the reviewers: 1. Dr Shriram K. Vasudevan (K. Ramakrishnan College of Technology) 2.
Maniroja M. Edinburgh (EXTC Dept., TSEC) 3. Nanda Dulal Jana (NIT, Durgapur) 4. Dipankar Dutta (NERIM Group of
Institutions) 5. Imthias Ahamed T P (TKM College of Engineering) 6. Dinobandhu Bhandari (Heritage Institute of
Technology) 7. Prof. BSP Mishra (KIIT, Patia) 8. Mrs Angeline (SRM University) 9. Dr Ram Mohan Dhara (IMT, Ghaziabad)
10. Saurabh Sharma (KIET Group of Institutions, Ghaziabad) 11. B. Lakshmanan (Mepco Schlenk Engg. College) 12. Dr
T. Subha (Sairam Engineering College) 13. Dr R. Ramya Devi (Velammal Engineering College) 14. Mr Zatin Kumar (Raj
Kumar Goel Institute of Technology, Ghaziabad) 15. M. Sowmiya (Easwari Engg. College) 16. A. Sekar (Knowledge
Institute of Tech.) ML_FM.indd 7 27­Mar­21 8:45:05 AM © Oxford University Press. All rights reserved. O x f o r d U n i v e
rsityPress
8. QR Code Content Details Scan the QR codes provided under the sections listed below to access the following
topics: Table of Contents • Appendix 1 – Python Language Fundamentals • Appendix 2 – Python Packages • Appendix
3 – Lab Manual (with 25 exercises) Chapter 1 • Section 1.4.4 – Important Machine Learning Algorithms Chapter 2 •
Section 2.1 – Machine Learning and Importance of Linear Algebra, Matrices and Tensors, Sampling Techniques,
Information Theory, Evaluation of Classifier Model and Additional Examples • Section 2.5 – Measures of Frequency
and Additional Examples • Exercises – Additional Key Terms, Short Questions and Numerical Problems Chapter 3 •
Section 3.1 – A figure showing Types of Learning • Section 3.4.4 – Additional Examples on Generalization and
Specialization • Section 3.4.6 – Additional Examples on Version Spaces Chapter 5 • Section 5.3 – Additional Examples
Chapter 6 • Section 6.2.1 – Additional Examples Chapter 7 • Section 7.1 – Simple OneR Algorithm • Section 7.8 –
Additional Examples • Section 7.8 – Additional Examples: Measure 3­Li몭 • Section 7.8 – Apriori Algorithm and
Frequent Pattern (FP) Growth Algorithm • Exercises – Additional Questions Chapter 8 • Section 8.1 – Probability
Theory and Additional Examples Chapter 9 • Section 9.3.1 – Additional Examples ML_FM.indd 8 27­Mar­21 8:45:05 AM
© Oxford University Press. All rights reserved. O x f o r d U n i v e r s i t y P r e s s
9. QR Code Content Details ix Chapter 10 • Section 10.1 – Convolution Neural Network, Modular Neural Network,
and Recurrent Neural Network • Key Terms – Additional Key Terms Chapter 11 • Section 11.1 – Decision Functions
and Additional Examples Chapter 12 • Section 12.2 – Illustrations of Types of Ensemble Models, Parallel Ensemble
Model, Voting, Bootstrap Resampling, Working Principle of Bagging, Learning Mechanism through Stacking, and
Sequential Ensemble Model • Section 12.4.1 – Gradient Tree Boosting and XGBoost • Exercises – Additional Review
Question Chapter 13 • Section 13.2 – Additional Information on Proximity Measures • Section 13.2 – Taxonomy of
Clustering Algorithms • Section 13.3.4 – Additional Examples • Section 13.4 – Additional Examples • Section 13.8 –
Purity, Evaluation based on Ground Truth, and Similarity­based Measures • Key Terms – Additional Key Terms •
Numerical Problems – Additional Numerical Problems Chapter 14 • Section 14.1 – Additional Information • Section
14.2 – Context of Reinforcement Learning and So몭Max Method • Section 14.7 – Additional Information on Dynamic
Programming • Section 14.7 – Additional Examples Chapter 15 • Section 15.2 – Examples of Optimization Problems
and Search Spaces • Section 15.3 – Flowchart of a Typical GA Algorithm • Section 15.4.6 – Additional Examples and
Information on Topics Discussed in Section 15.4 • Section 15.5.1 – Feature Selection using Genetic Algorithms and
Additional Content on Genetic Algorithm Classifier • Section 15.6.2 – Additional Information on Genetic Programming
Chapter 16 • Section 16.2 – Additional Information on Activation Functions • Section 16.3 – L1 and L2 Regularizations •
Section 16.4 – Additional Information on Input Layer • Section 16.4 – Additional Information on Padding • Section
16.5 – Detailed Information on Transfer Learning • Section 16.8 – Restricted Boltzmann Machines and Deep Belief
Networks, Auto Encoders, and Deep Reinforcement Learning • Exercises – Additional Questions ML_FM.indd 9 27­
Mar­21 8:45:05 AM © Oxford University Press. All rights reserved. O x f o r d U n i v e r s i t y P r e s s
10. Detailed Contents Preface iv Acknowledgements vii QR Code Content Details viii 1. Introduction to Machine
Learning1 1.1 Need for Machine Learning 1 1.2 Machine Learning Explained 3 1.3 Machine Learning in Relation to
Other Fields 5 1.3.1 Machine Learning and Artificial Intelligence 5 1.3.2 Machine Learning, Data Science, Data Mining,
and Data Analytics 5 1.3.3 Machine Learning and Statistics6 1.4 Types of Machine Learning 7 1.4.1 Supervised
Learning 8 1.4.2 Unsupervised Learning 11 1.4.3 Semi­supervised Learning 12 1.4.4 Reinforcement Learning 12 1.5
Challenges of Machine Learning 13 1.6 Machine Learning Process 14 1.7 Machine Learning Applications 15 2.
Understanding Data 22 2.1 What is Data? 22 2.1.1 Types of Data 24 2.1.2 Data Storage and Representation25 2.2 Big
Data Analytics and Types of Analytics26 2.3 Big Data Analysis Framework 27 2.3.1 Data Collection 29 2.3.2 Data
Preprocessing 30 2.4 Descriptive Statistics 34 2.5 Univariate Data Analysis and Visualization36 2.5.1 Data Visualization
36 2.5.2 Central Tendency 38 2.5.3 Dispersion 40 2.5.4 Shape 41 2.5.5 Special Univariate Plots 43 2.6 Bivariate Data
and Multivariate Data44 2.6.1 Bivariate Statistics 46 2.7 Multivariate Statistics 47 2.8 Essential Mathematics for
Multivariate Data 49 2.8.1 Linear Systems and Gaussian Elimination for Multivariate Data49 2.8.2 Matrix
Decompositions 50 2.8.3 Machine Learning and Importance of Probability and Statistics 52 2.9 Overview of
Hypothesis 57 2.9.1 Comparing Learning Methods59 2.10 Feature Engineering and Dimensionality Reduction
Techniques62 2.10.1 Stepwise Forward Selection63 2.10.2 Stepwise Backward Elimination63 2.10.3 Principal
Component Analysis63 2.10.4 Linear Discriminant Analysis67 2.10.5 Singular Value Decomposition68 ML_FM.indd 10
27­Mar­21 8:45:05 AM © Oxford University Press. All rights reserved. O x f o r d U n i v e r s i t y P r e s s
11. Detailed Contents xi 3. Basics of Learning Theory 77 3.1 Introduction to Learning and its Types 78 3.2
Introduction to Computation Learning Theory 80 3.3 Design of a Learning System 81 3.4 Introduction to Concept
Learning82 3.4.1 Representation of a Hypothesis83 3.4.2 Hypothesis Space 85 3.4.3 Heuristic Space Search 85 3.4.4
Generalization and Specialization86 3.4.5 Hypothesis Space Search by Find­S Algorithm 88 3.4.6 Version Spaces 90
3.5 Induction Biases 94 3.5.1 Bias and Variance 95 3.5.2 Bias vs Variance Tradeo몭 96 3.5.3 Best Fit in Machine Learning
97 3.6 Modelling in Machine Learning 97 3.6.1 Model Selection and Model Evaluation98 3.6.2 Re­sampling Methods 99
3.7 Learning Frameworks 104 3.7.1 PAC Framework 104 3.7.2 Estimating Hypothesis Accuracy106 3.7.3 Hoe몭ding’s
Inequality 106 3.7.4 Vapnik–Chervonenkis Dimension107 4. Similarity­based Learning 115 4.1 Introduction to
Similarity or Instance­based Learning 116 4.1.1 Di몭erences Between Instance­ and Model­based Learning 116 4.2
Nearest­Neighbor Learning 117 4.3 Weighted K­Nearest­Neighbor Algorithm120 4.4 Nearest Centroid Classifier 123 4.5
Locally Weighted Regression (LWR) 124 5. Regression Analysis 130 5.1 Introduction to Regression 130 5.2 Introduction
to Linearity, Correlation, and Causation 131 5.3 Introduction to Linear Regression134 5.4 Validation of Regression
Methods138 5.5 Multiple Linear Regression 141 5.6 Polynomial Regression 142 5.7 Logistic Regression 144 5.8 Ridge,
Lasso, and Elastic Net Regression 147 5.8.1 Ridge Regularization 148 5.8.2 LASSO 149 5.8.3 Elastic Net 149 6. Decision
Tree Learning 155 6.1 Introduction to Decision Tree Learning Model 155 6.1.1 Structure of a Decision Tree156 6.1.2
Fundamentals of Entropy 159 6.2 Decision Tree Induction Algorithms161 6.2.1 ID3 Tree Construction 161 6.2.2 C4.5
Construction 167 6.2.3 Classification and Regression Trees Construction 175 6.2.4 Regression Trees 185 6.3 Validating
and Pruning of Decision Trees 190 ML_FM.indd 11 27­Mar­21 8:45:05 AM © Oxford University Press. All rights reserved.
OxfordUniversityPress
12. xii Detailed Contents 7. Rule–based Learning 196 7.1 Introduction 196 7.2 Sequential Covering Algorithm 198
7.2.1 PRISM 198 7.3 First Order Rule Learning 206 7.3.1 FOIL (First Order Inductive Learner Algorithm) 208 7.4 Induction
as Inverted Deduction 215 7.5 Inverting Resolution 215 7.5.1 Resolution Operator (Propositional Form) 215 7.5.2
Inverse Resolution Operator (Propositional Form) 216 7.5.3 First Order Resolution 216 7.5.4 Inverting First Order
Resolution216 7.6 Analytical Learning or Explanation Based Learning (EBL) 217 7.6.1 Perfect Domain Theories 218 7.7
Active Learning 221 7.7.1 Active Learning Mechanisms222 7.7.2 Query Strategies/Selection Strategies223 7.8
Association Rule Mining 225 8. Bayesian Learning 234 8.1 Introduction to Probability­based Learning234 8.2
Fundamentals of Bayes Theorem 235 8.3 Classification Using Bayes Model 235 8.3.1 Naïve Bayes Algorithm 237 8.3.2
Brute Force Bayes Algorithm243 8.3.3 Bayes Optimal Classifier 243 8.3.4 Gibbs Algorithm 244 8.4 Naïve Bayes
Algorithm for Continuous Attributes 244 8.5 Other Popular Types of Naive Bayes Classifiers 247 9. Probabilistic
Graphical Models 253 9.1 Introduction 253 9.2 Bayesian Belief Network 254 9.2.1 Constructing BBN 254 9.2.2 Bayesian
Inferences 256 9.3 Markov Chain 261 9.3.1 Markov Model 261 9.3.2 Hidden Markov Model 263 9.4 Problems Solved
with HMM 264 9.4.1 Evaluation Problem 265 9.4.2 Computing Likelihood Probability267 9.4.3 Decoding Problem 269
9.4.4 Baum­Welch Algorithm 272 10. Artificial Neural Networks 279 10.1 Introduction 280 10.2 Biological Neurons 280
10.3 Artificial Neurons 281 10.3.1 Simple Model of an Artificial Neuron281 10.3.2 Artificial Neural Network Structure282
10.3.3 Activation Functions 282 10.4 Perceptron and Learning Theory284 10.4.1 XOR Problem 287 10.4.2 Delta
Learning Rule and Gradient Descent 288 10.5 Types of Artificial Neural Networks288 10.5.1 Feed Forward Neural
Network289 10.5.2 Fully Connected Neural Network289 10.5.3 Multi­Layer Perceptron (MLP)289 10.5.4 Feedback
Neural Network290 10.6 Learning in a Multi­Layer Perceptron290 ML_FM.indd 12 27­Mar­21 8:45:05 AM © Oxford
University Press. All rights reserved. O x f o r d U n i v e r s i t y P r e s s
13. Detailed Contents xiii 10.7 Radial Basis Function Neural Network297 10.8 Self­Organizing Feature Map 301 10.9
Popular Applications of Artificial Neural Networks 306 10.10 Advantages and Disadvantages of ANN 306 10.11
Challenges of Artificial Neural Networks307 11. Support Vector Machines 312 11.1 Introduction to Support Vector
Machines 312 11.2 Optimal Hyperplane 314 11.3 Functional and Geometric Margin316 11.4 Hard Margin SVM as an
Optimization Problem 319 11.4.1 Lagrangian Optimization Problem320 11.5 So몭 Margin Support Vector Machines323
11.6 Introduction to Kernels and Non­Linear SVM 326 11.7 Kernel­based Non­Linear Classifier 330 11.8 Support Vector
Regression 331 11.8.1 Relevance Vector Machines333 12. Ensemble Learning 339 12.1 Introduction 339 12.1.1
Ensembling Techniques 341 12.2 Parallel Ensemble Models 341 12.2.1 Voting 341 12.2.2 Bootstrap Resampling 341
12.2.3 Bagging 342 12.2.4 Random Forest 342 12.3 Incremental Ensemble Models 346 12.3.1 Stacking 347 12.3.2
Cascading 347 12.4 Sequential Ensemble Models 347 12.4.1 AdaBoost 347 13. Clustering Algorithms 361 13.1
Introduction to Clustering Approaches361 13.2 Proximity Measures 364 13.3 Hierarchical Clustering Algorithms368
13.3.1 Single Linkage or MIN Algorithm368 13.3.2 Complete Linkage or MAX or Clique 371 13.3.3 Average Linkage 371
13.3.4 Mean­Shi몭 Clustering Algorithm372 13.4 Partitional Clustering Algorithm 373 13.5 Density­based Methods 376
13.6 Grid­based Approach 377 13.7 Probability Model­based Methods379 13.7.1 Fuzzy Clustering 379 13.7.2
Expectation­Maximization (EM) Algorithm 380 13.8 Cluster Evaluation Methods 382 14. Reinforcement Learning 389
14.1 Overview of Reinforcement Learning389 14.2 Scope of Reinforcement Learning390 14.3 Reinforcement Learning
As Machine Learning 392 14.4 Components of Reinforcement Learning393 14.5 Markov Decision Process 396 14.6
Multi­Arm Bandit Problem and Reinforcement Problem Types398 ML_FM.indd 13 27­Mar­21 8:45:05 AM © Oxford
University Press. All rights reserved. O x f o r d U n i v e r s i t y P r e s s
14. xiv Detailed Contents 14.7 Model­based Learning (Passive Learning)402 14.8 Model Free Methods 406 14.8.1
Monte­Carlo Methods 407 14.8.2 Temporal Di몭erence Learning408 14.9 Q­Learning409 14.10 SARSA Learning 410 15.
Genetic Algorithms 417 15.1 Overview of Genetic Algorithms417 15.2 Optimization Problems and Search Spaces 419
15.3 General Structure of a Genetic Algorithm 420 15.4 Genetic Algorithm Components 422 15.4.1 Encoding Methods
422 15.4.2 Population Initialization 424 15.4.3 Fitness Functions 424 15.4.4 Selection Methods 425 15.4.5 Crossover
Methods 428 15.4.6 Mutation Methods 429 15.5 Case Studies in Genetic Algorithms430 15.5.1 Maximization of a
Function430 15.5.2 Genetic Algorithm Classifier 433 15.6 Evolutionary Computing 433 15.6.1 Simulated Annealing 433
15.6.2 Genetic Programming 434 16. Deep Learning 439 16.1 Introduction to Deep Neural Networks439 16.2
Introduction to Loss Functions and Optimization 440 16.3 Regularization Methods 442 16.4 Convolutional Neural
Networks 444 16.5 Transfer Learning 451 16.6 Applications of Deep Learning 451 16.6.1 Robotic Control 451 16.6.2
Linear Systems and Non­linear Dynamics 452 16.6.3 Data Mining 452 16.6.4 Autonomous Navigation 453 16.6.5
Bioinformatics 453 16.6.6 Speech Recognition 453 16.6.7 Text Analysis 454 16.7 Recurrent Neural Networks 454 16.8
LSTM and GRU 457 Bibliography463 Index472 About the Authors 480 Related Titles 481 Scan for ‘Appendix 1 ­ Python
Language Fundamentals' Scan for ‘Appendix 2 ­ Python Packages' Scan for ‘Appendix 3 ­ Lab Manual with 25
Exercises' ML_FM.indd 14 27­Mar­21 8:45:06 AM © Oxford University Press. All rights reserved. O x f o r d U n i v e r s i t y
Press
15. Introduction to Machine Learning Chapter 1 1.1 NEED FOR MACHINE LEARNING Business organizations use huge
amount of data for their daily activities. Earlier, the full potential of this data was not utilized due to two reasons. One
reason was data being scattered across di몭erent archive systems and organizations not being able to integrate these
sources fully. Secondly, the lack of awareness about so몭ware tools that could help to unearth the useful information
from data. Not anymore! Business organizations have now started to use the latest technology, machine learning, for
this purpose. Machine learning has become so popular because of three reasons: 1. High volume of available data to
manage: Big companies such as Facebook, Twitter, and YouTube generate huge amount of data that grows at a
phenomenal rate. It is estimated that the data approximately gets doubled every year. Machine Learning (ML) is a
promising and flourishing field. It can enable top management of an organization to extract the knowledge from the
data stored in various archives of the business organizations to facilitate decision making. Such decisions can be
useful for organizations to design new products, improve business processes, and to develop decision support
systems. Learning Objectives • Explore the basics of machine learning • Introduce types of machine learning • Provide
an overview of machine learning tasks • State the components of the machine learning algorithm • Explore the
machine learning process • Survey some machine learning applications “Computers are able to see, hear and learn.
Welcome to the future.” — Dave Waters ML_01.indd 1 26­Mar­21 3:12:40 PM © Oxford University Press. All rights
reserved. O x f o r d U n i v e r s i t y P r e s s
16. 2 Machine Learning 2. Second reason is that the cost of storage has reduced. The hardware cost has also
dropped. Therefore, it is easier now to capture, process, store, distribute, and transmit the digital information. 3. Third
reason for popularity of machine learning is the availability of complex algorithms now. Especially with the advent of
deep learning, many algorithms are available for machine learning. With the popularity and ready adaption of
machine learning by business organizations, it has become a dominant technology trend now. Before starting the
machine learning journey, let us establish these terms ­ data, information, knowledge, intelligence, and wisdom. A
knowledge pyramid is shown in Figure 1.1. Wisdom Intelligence (applied knowledge) Knowledge (condensed
information) Information (processed data) Data (mostly available as raw facts and symbols) Figure 1.1: The
Knowledge Pyramid What is data? All facts are data. Data can be numbers or text that can be processed by a
computer. Today, organizations are accumulating vast and growing amounts of data with data sources such as flat
files, databases, or data warehouses in di몭erent storage formats. Processed data is called information. This includes
patterns, associations, or relationships among data. For example, sales data can be analyzed to extract information
like which is the fast selling product. Condensed information is called knowledge. For example, the historical patterns
and future trends obtained in the above sales data can be called knowledge. Unless knowledge is extracted, data is
of no use. Similarly, knowledge is not useful unless it is put into action. Intelligence is the applied knowledge for
actions. An actionable form of knowledge is called intelligence. Computer systems have been successful till this
stage. The ultimate objective of knowledge pyramid is wisdom that represents the maturity of mind that is, so far,
exhibited only by humans. Here comes the need for machine learning. The objective of machine learning is to
process these archival data for organizations to take better decisions to design new products, improve the business
processes, and to develop e몭ective decision support systems. ML_01.indd 2 26­Mar­21 3:12:40 PM © Oxford
University Press. All rights reserved. O x f o r d U n i v e r s i t y P r e s s
17. Introduction to Machine Learning 3 1.2 MACHINE LEARNING EXPLAINED Machine learning is an important sub­
branch of Artificial Intelligence (AI). A frequently quoted definition of machine learning was by Arthur Samuel, one of
the pioneers of Artificial Intelligence. He stated that “Machine learning is the field of study that gives the computers
ability to learn without being explicitly programmed.”
Thekeytothisdefinitionisthatthesystemsshouldlearnbyitselfwithoutexplicitprogramming. How is it possible? It is
widely known that to perform a computation, one needs to write programs that teach the computers how to do that
computation. In conventional programming, a몭er understanding the problem, a detailed design of the program such
as a flowchart or an algorithm needs to be created and converted into programs using a suitable programming
language. This approach could be di몭icult for many real­world problems such as puzzles, games, and complex image
recognition applications. Initially, artificial intelligence aims to understand these problems and develop general
purpose rules manually. Then, these rules are formulated into logic and implemented in a program to create
intelligent systems. This idea of developing intelligent systems by using logic and reasoning by converting an expert’s
knowledge into a set of rules and programs is called an expert system. An expert system like MYCIN was designed for
medical diagnosis a몭er converting the expert knowledge of many doctors into a system. However, this approach did
not progress much as programs lacked real intelligence. The word MYCIN is derived from the fact that most of the
antibiotics’ names end with ‘mycin’. The above approach was impractical in many domains as programs still
depended on human expertise and hence did not truly exhibit intelligence. Then, the momentum shi몭ed to machine
learning in the form of data driven systems. The focus of AI is to develop intelligent systems by using data­driven
approach, where data is used as an input to develop intelligent models. The models can then be used to predict new
inputs. Thus, the aim of machine learning is to learn a model or set of rules from the given dataset automatically so
that it can predict the unknown data correctly. As humans take decisions based on an experience, computers make
models based on extracted patterns in the input data and then use these data­filled models for prediction and to
take decisions. For computers, the learnt model is equivalent to human experience. This is shown in Figure 1.2.
Experience Humans (a) (b) Decisions Data Data­ base Learning program Model Figure 1.2: (a) A Learning System for
Humans (b) A Learning System for Machine Learning O몭en, the quality of data determines the quality of experience
and, therefore, the quality of the learning system. In statistical learning, the relationship between the input x and
output y is ML_01.indd 3 26­Mar­21 3:12:40 PM © Oxford University Press. All rights reserved. O x f o r d U n i v e r s i t y
Press
18. 4 Machine Learning modeled as a function in the form y = f(x). Here, f is the learning function that maps the
input x to output y. Learning of function f is the crucial aspect of forming a model in statistical learning. In machine
learning, this is simply called mapping of input to output. The learning program summarizes the raw data in a model.
Formally stated, a model is an explicit description of patterns within the data in the form of: 1. Mathematical equation
2. Relational diagrams like trees/graphs 3. Logical if/else rules, or 4. Groupings called clusters In summary, a model
can be a formula, procedure or representation that can generate data decisions. The di몭erence between pattern and
model is that the former is local and applicable only to certain attributes but the latter is global and fits the entire
dataset. For example, a model can be helpful to examine whether a given email is spam or not. The point is that the
model is generated automatically from the given data. Another pioneer of AI, Tom Mitchell’s definition of machine
learning states that, “A computer program is said to learn from experience E, with respect to task T and some
performance measure P, if its performance on T measured by P improves with experience E.” The important
components of this definition are experience E, task T, and performance measure P. For example, the task T could be
detecting an object in an image. The machine can gain the knowledge of object using training dataset of thousands
of images. This is called experience E. So, the focus is to use this experience E for this task of object detection T. The
ability of the system to detect the object is measured by performance measures like precision and recall. Based on
the performance measures, course correction can be done to improve the performance of the system. Models of
computer systems are equivalent to human experience. Experience is based on data. Humans gain experience by
various means. They gain knowledge by rote learning. They observe
othersandimitateit.Humansgainalotofknowledgefromteachersandbooks.Welearnmanythings by trial and error. Once
the knowledge is gained, when a new problem is encountered, humans search for similar past situations and then
formulate the heuristics and use that for prediction. But, in systems, experience is gathered by these steps: 1.
Collection of data 2. Once data is gathered, abstract concepts are formed out of that data. Abstraction is used to
generate concepts. This is equivalent to humans’ idea of objects, for example, we have some idea about how an
elephant looks like. 3. Generalization converts the abstraction into an actionable form of intelligence. It can be
viewed as ordering of all possible concepts. So, generalization involves ranking of concepts, inferencing from them
and formation of heuristics, an actionable aspect of intelligence. Heuristics are educated guesses for all tasks. For
example, if one runs or encounters a danger, it is the resultant of human experience or his heuristics formation. In
machines, it happens the same way. 4. Heuristics normally works! But, occasionally, it may fail too. It is not the fault
of heuristics as it is just a ‘rule of thumb′. The course correction is done by taking evaluation measures. Evaluation
checks the thoroughness of the models and to­do course correction, if necessary, to generate better formulations.
ML_01.indd 4 26­Mar­21 3:12:40 PM © Oxford University Press. All rights reserved. O x f o r d U n i v e r s i t y P r e s s
19. Introduction to Machine Learning 5 1.3 MACHINE LEARNING IN RELATION TO OTHER FIELDS Machine learning
uses the concepts of Artificial Intelligence, Data Science, and Statistics primarily. It is the resultant of combined ideas
of diverse fields. 1.3.1 Machine Learning and Artificial Intelligence Machine learning is an important branch of AI,
which is a much broader subject. The aim of AI is to develop intelligent agents. An agent can be a robot, humans, or
any autonomous systems. Initially, the idea of AI was ambitious, that is, to develop intelligent systems like human
beings. The focus was on logic and logical inferences. It had seen many ups and downs. These down periods were
called AI winters. The resurgence in AI happened due to development of data driven systems. The aim is to find
relations and regularities present in the data. Machine learning is the subbranch of AI, whose aim is to extract the
patterns for prediction. It is a broad field that includes learning from examples and other areas like reinforcement
learning. The relationship of AI and machine learning is shown in Figure 1.3. The model can take an unknown
instance and generate results. Artificial intelligence Machine learning Deep learning Figure 1.3: Relationship of AI with
Machine Learning Deep learning is a subbranch of machine learning. In deep learning, the models are constructed
using neural network technology. Neural networks are based on the human neuron models. Many neurons form a
network connected with the activation functions that trigger further neurons to perform tasks. 1.3.2 Machine
Learning, Data Science, Data Mining, and Data Analytics Data science is an ‘Umbrella’ term that encompasses many
fields. Machine learning starts with data. Therefore, data science and machine learning are interlinked. Machine
learning is a branch of data science. Data science deals with gathering of data for analysis. It is a broad field that
includes: ML_01.indd 5 26­Mar­21 3:12:40 PM © Oxford University Press. All rights reserved. O x f o r d U n i v e r s i t y P
ress
20. 6 Machine Learning Big Data Data science concerns about collection of data. Big data is a field of data science
that deals with data’s following characteristics: 1. Volume: Huge amount of data is generated by big companies like
Facebook, Twitter, YouTube. 2. Variety: Data is available in variety of forms like images, videos, and in di몭erent
formats. 3. Velocity: It refers to the speed at which the data is generated and processed. Big data is used by many
machine learning algorithms for applications such as language trans­ lation and image recognition. Big data
influences the growth of subjects like Deep learning. Deep learning is a branch of machine learning that deals with
constructing models using neural networks. Data Mining Data mining’s original genesis is in the business. Like while
mining the earth one gets into precious resources, it is o몭en believed that unearthing of the data produces hidden
infor­ mation that otherwise would have eluded the attention of the management. Nowadays, many consider that
data mining and machine learning are same. There is no di몭erence between these fields except that data mining
aims to extract the hidden patterns that are present in the data, whereas, machine learning aims to use it for
prediction. Data Analytics Another branch of data science is data analytics. It aims to extract useful knowledge from
crude data. There are di몭erent types of analytics. Predictive data analytics is used for making predictions. Machine
learning is closely related to this branch of analytics and shares almost all algorithms. Pattern Recognition It is an
engineering field. It uses machine learning algorithms to extract the features for pattern analysis and pattern
classification. One can view pattern recognition as a specific application of machine learning. These relations are
summarized in Figure 1.4. Data science Data mining Machine learning Pattern recognition Big data Data analytics
Figure 1.4: Relationship of Machine Learning with Other Major Fields 1.3.3 Machine Learning and Statistics
Statisticsisabranchofmathematicsthathasasolidtheoreticalfoundationregardingstatisticallearning. Like machine
learning (ML), it can learn from data. But the di몭erence between statistics and ML is that statistical methods look for
regularity in data called patterns. Initially, statistics sets a hypothesis and performs experiments to verify and validate
the hypothesis in order to find relationships among data. ML_01.indd 6 26­Mar­21 3:12:41 PM © Oxford University
Press. All rights reserved. O x f o r d U n i v e r s i t y P r e s s
21. Introduction to Machine Learning 7 Statistics requires knowledge of the statistical procedures and the guidance
of a good statistician. It is mathematics intensive and models are o몭en complicated equations and involve many
assumptions. Statistical methods are developed in relation to the data being analysed. In addition, statistical
methods are coherent and rigorous. It has strong theoretical foundations and interpretations that require a strong
statistical knowledge. Machine learning, comparatively, has less assumptions and requires less statistical knowledge.
But, it o몭en requires interaction with various tools to automate the process of learning. Nevertheless, there is a
school of thought that machine learning is just the latest version of ‘old Statistics’ and hence this relationship should
be recognized. 1.4 TYPES OF MACHINE LEARNING What does the word ‘learn’ mean? Learning, like adaptation, occurs
as the result of interaction of the program with its environment. It can be compared with the interaction between a
teacher and a student. There are four types of machine learning as shown in Figure 1.5. Machine learning Supervised
learning Regression Cluster analysis Association mining Dimension reduction Unsupervised learning Semi­
supervised learning Reinforcement learning Figure 1.5: Types of Machine Learning Before discussing the types of
learning, it is necessary to discuss about data. Labelled and Unlabelled Data Data is a raw fact. Normally, data is
represented in the form of a table. Data also can be referred to as a data point, sample, or an example. Each row of
the table represents a data point. Features are attributes or characteristics of an object. Normally, the columns of the
table are attributes. Out of all attributes, one attribute is important and is called a label. Label is the feature that we
aim to predict. Thus, there are two types of data – labelled and unlabelled. Labelled Data To illustrate labelled data,
let us take one example dataset called Iris flower dataset or Fisher’s Iris dataset. The dataset has 50 samples of Iris –
with four attributes, length and width of sepals and petals. The target variable is called class. There are three classes –
Iris setosa, Iris virginica, and Iris versicolor. The partial data of Iris dataset is shown in Table 1.1. ML_01.indd 7 26­Mar­
21 3:12:41 PM © Oxford University Press. All rights reserved. O x f o r d U n i v e r s i t y P r e s s
22. 8 Machine Learning Table 1.1: Iris Flower Dataset S.No. Length of Petal Width of Petal Length of Sepal Width of
Sepal Class 1. 5.5 4.2 1.4 0.2 Setosa 2. 7 3.2 4.7 1.4 Versicolor 3. 7.3 2.9 6.3 1.8 Virginica A dataset need not be always
numbers. It can be images or video frames. Deep neural networks can handle images with labels. In the following
Figure 1.6, the deep neural network takes images of dogs and cats with labels for classification. Input Label dog Cat
(a) (b) Figure 1.6: (a) Labelled Dataset (b) Unlabelled Dataset In unlabelled data, there are no labels in the dataset.
1.4.1 Supervised Learning Supervised algorithms use labelled dataset. As the name suggests, there is a supervisor or
teacher component in supervised learning. A supervisor provides labelled data so that the model is constructed and
generates test data. In supervised learning algorithms, learning takes place in two stages. In layman terms, during the
first stage, the teacher communicates the information to the student that the student is supposed to master. The
student receives the information and understands it. During this stage, the teacher has no knowledge of whether the
information is grasped by the student. This leads to the second stage of learning. The teacher then asks the student a
set of questions to find out how much information has been grasped by the student. Based on these questions,
ML_01.indd 8 26­Mar­21 3:12:41 PM © Oxford University Press. All rights reserved. O x f o r d U n i v e r s i t y P r e s s
23. Introduction to Machine Learning 9 the student is tested, and the teacher informs the student about his
assessment. This kind of learning is typically called supervised learning. Supervised learning has two methods: 1.
Classification 2. Regression Classification Classification is a supervised learning method. The input attributes of the
classification algorithms are called independent variables. The target attribute is called label or dependent variable.
The relationship between the input and target variable is represented in the form of a structure which is called a
classification model. So, the focus of classification is to predict the ‘label’ that is in a discrete form (a value from the
set of finite values). An example is shown in Figure 1.7 where a classification algorithm takes a set of labelled data
images such as dogs and cats to construct a model that can later be used to classify an unknown test image data.
Labelled data Classification algorithm Classification model New test data Label is Cat Figure 1.7: An Example
Classification System In classification, learning takes place in two stages. During the first stage, called training stage,
the learning algorithm takes a labelled dataset and starts learning. A몭er the training set, samples are processed and
the model is generated. In the second stage, the constructed model is tested with test or unknown sample and
assigned a label. This is the classification process. This is illustrated in the above Figure 1.7. Initially, the classification
learning algorithm learns with the collection of labelled data and constructs the model. Then, a test case is selected,
and the model assigns a label. Similarly, in the case of Iris dataset, if the test is given as (6.3, 2.9, 5.6, 1.8, ?), the
classification will generate the label for this. This is called classification. One of the examples of classification is –
Image recognition, which includes classification of diseases like cancer, classification of plants, etc. The classification
models can be categorized based on the implementation technology like
decisiontrees,probabilisticmethods,distancemeasures,andso몭computingmethods.Classification models can also be
classified as generative models and discriminative models. Generative models deal with the process of data
generation and its distribution. Probabilistic models are examples of ML_01.indd 9 26­Mar­21 3:12:41 PM © Oxford
University Press. All rights reserved. O x f o r d U n i v e r s i t y P r e s s
24. 10 Machine Learning generative models. Discriminative models do not care about the generation of data.
Instead, they simply concentrate on classifying the given data. Some of the key algorithms of classification are: •
Decision Tree • Random Forest • Support Vector Machines • Naïve Bayes • Artificial Neural Network and Deep Learning
networks like CNN Regression Models Regression models, unlike classification algorithms, predict continuous
variables like price. In other words, it is a number. A fitted regression model is shown in Figure 1.8 for a dataset that
represent weeks input x and product sales y. 1 1 1.5 y­axis ­ Product sales data (y) 2 2.5 3 3.5 4 2 x­axis ­ Week data (x)
Regression line (y = 0.66X + 0.54) 3 4 5 Figure 1.8: A Regression Model of the Form y = ax + b The regression model
takes input x and generates a model in the form of a fitted line of the form y = f(x). Here, x is the independent variable
that may be one or more attributes and y is the dependent variable. In Figure 1.8, linear regression takes the training
set and tries to fit it with a line – product sales = 0.66 × Week + 0.54. Here, 0.66 and 0.54 are all regression coe몭icients
that are learnt from data. The advantage of this model is that prediction for product sales (y) can be made for
unknown week data (x). For example, the prediction for unknown eighth week can be made by substituting x as 8 in
that regression formula to get y. One of the most important regression algorithms is linear regression that is
explained in the next section. Both regression and classification models are supervised algorithms. Both have a
supervisor and the concepts of training and testing are applicable to both. What is the di몭erence between
classification and regression models? The main di몭erence is that regression models predict continuous variables
such as product price, while classification concentrates on assigning labels such as class. ML_01.indd 10 26­Mar­21
3:12:41 PM © Oxford University Press. All rights reserved. O x f o r d U n i v e r s i t y P r e s s
25. Introduction to Machine Learning 11 1.4.2 Unsupervised Learning The second kind of learning is by self­
instruction. As the name suggests, there are no supervisor or teacher components. In the absence of a supervisor or
teacher, self­instruction is the most common kind of learning process. This process of self­instruction is based on the
concept of trial and error. Here, the program is supplied with objects, but no labels are defined. The algorithm itself
observes the examples and recognizes patterns based on the principles of grouping. Grouping is done in ways that
similar objects form the same group. Cluster analysis and Dimensional reduction algorithms are examples of
unsupervised algorithms. Cluster Analysis Cluster analysis is an example of unsupervised learning. It aims to group
objects into disjoint clusters or groups. Cluster analysis clusters objects based on its attributes. All the data objects of
the partitions are similar in some aspect and vary from the data objects in the other partitions significantly. Some of
the examples of clustering processes are — segmentation of a region of interest in an image, detection of abnormal
growth in a medical image, and determining clusters of signatures in a gene database. An example of clustering
scheme is shown in Figure 1.9 where the clustering algorithm takes a set of dogs and cats images and groups it as
two clusters­dogs and cats. It can be observed that the samples belonging to a cluster are similar and samples are
di몭erent radically across clusters. Unlabelled data Clustering algorithm Cluster 1 Cluster 2 Figure 1.9: An Example
Clustering Scheme Some of the key clustering algorithms are: • k­means algorithm • Hierarchical algorithms
ML_01.indd 11 26­Mar­21 3:12:41 PM © Oxford University Press. All rights reserved. O x f o r d U n i v e r s i t y P r e s s
26. 12 Machine Learning Dimensionality Reduction Dimensionality reduction algorithms are examples of
unsupervised algorithms. It takes a higher dimension data as input and outputs the data in lower dimension by
taking advantage of the variance of the data. It is a task of reducing the dataset with few features without losing the
generality. The di몭erences between supervised and unsupervised learning are listed in the following Table 1.2. Table
1.2: Di몭erences between Supervised and Unsupervised Learning S.No. Supervised Learning Unsupervised Learning 1.
There is a supervisor component No supervisor component 2. Uses Labelled data Uses Unlabelled data 3. Assigns
categories or labels Performs grouping process such that similar objects will be in one cluster 1.4.3 Semi­supervised
Learning There are circumstances where the dataset has a huge collection of unlabelled data and some labelled
data. Labelling is a costly process and di몭icult to perform by the humans. Semi­supervised algorithms use unlabelled
data by assigning a pseudo­label. Then, the labelled and pseudo­labelled dataset can be combined. 1.4.4
Reinforcement Learning Reinforcement learning mimics human beings. Like human beings use ears and eyes to
perceive the world and take actions, reinforcement learning allows the agent to interact with the environment to get
rewards. The agent can be human, animal, robot, or any independent program. The rewards enable the agent to gain
experience. The agent aims to maximize the reward. The reward can be positive or negative (Punishment). When the
rewards are more, the behavior gets reinforced and learning becomes possible. Consider the following example of a
Grid game as shown in Figure 1.10. Block Goal Danger Figure 1.10: A Grid game ML_01.indd 12 26­Mar­21 3:12:41 PM
© Oxford University Press. All rights reserved. O x f o r d U n i v e r s i t y P r e s s
27. Introduction to Machine Learning 13 1.5 CHALLENGES OF MACHINE LEARNING What are the challenges of
machine learning? Let us discuss about them now. Problems that can be Dealt with Machine Learning
Computersarebetterthanhumansinperformingtaskslikecomputation.Forexample,whilecalculating the square root of
large numbers, an average human may blink but computers can display the result in seconds. Computers can play
games like chess, GO, and even beat professional players of that game. However, humans are better than computers
in many aspects like recognition. But, deep learning systems challenge human beings in this aspect as well.
Machines can recognize human faces in a second. Still, there are tasks where humans are better as machine learning
systems still require quality data for model construction. The quality of a learning system depends on the quality of
data. This is a challenge. Some of the challenges are listed below: 1. Problems–Machinelearningcandealwiththe‘well­
posed’problemswherespecifications are complete and available. Computers cannot solve ‘ill­posed’ problems.
Consider one simple example (shown in Table 1.3): Table 1.3: An Example Input (x1 , x2 ) Output (y) 1, 1 1 2, 1 2 3, 1 3 4,
1 4 5, 1 5 Can a model for this test data be multiplication? That is, y = x1 × x2 . Well! It is true! But, this is equally true
that y may be y = x1 ÷ x2 , or y = x1 x2. So, there are three functions that fit the data. This means that the problem is ill­
posed. To solve this problem, one needs more example to check the model. Puzzles and games that do not have
su몭icient specification may become an ill­posed problem and scientific computation has many ill­posed problems.
Scan for information on ‘Important Machine Learning Algorithms’ In this grid game, the gray tile indicates the danger,
black is a block, and the tile with diagonal lines is the goal. The aim is to start, say from bottom­le몭 grid, using the
actions le몭, right, top and bottom to reach the goal state. To solve this sort of problem, there is no data. The agent
interacts with the environment to get experience. In the above case, the agent tries to create a model by simulating
many paths and finding rewarding paths. This experience helps in constructing a model. It can be said in summary,
compared to supervised learning, there is no supervisor or labelled dataset. Many sequential decisions need to be
taken to reach the final decision. Therefore, reinforcement algorithms are reward­based, goal­oriented algorithms.
ML_01.indd 13 26­Mar­21 3:12:41 PM © Oxford University Press. All rights reserved. O x f o r d U n i v e r s i t y P r e s s
28. 14 Machine Learning 2. Huge data – This is a primary requirement of machine learning. Availability of a quality
data is a challenge. A quality data means it should be large and should not have data problems such as missing data
or incorrect data. 3. High computation power – With the availability of Big Data, the computational resource
requirement has also increased. Systems with Graphics Processing Unit (GPU) or even Tensor Processing Unit (TPU)
are required to execute machine learning algorithms. Also, machine learning tasks have become complex and hence
time complexity has increased, and that can be solved only with high computing power. 4. Complexity of the
algorithms – The selection of algorithms, describing the algorithms, application of algorithms to solve machine
learning task, and comparison of algorithms have become necessary for machine learning or data scientists now.
Algorithms have become a big topic of discussion and it is a challenge for machine learning professionals to design,
select, and evaluate optimal algorithms. 5. Bias/Variance – Variance is the error of the model. This leads to a problem
called bias/ variance tradeo몭. A model that fits the training data correctly but fails for test data, in general lacks
generalization, is called overfitting. The reverse problem is called underfitting where the model fails for training data
but has good generalization. Overfitting and underfitting are great challenges for machine learning algorithms. 1.6
MACHINE LEARNING PROCESS The emerging process model for the data mining solutions for business organizations
is CRISP­DM. Since machine learning is like data mining, except for the aim, this process can be used for machine
learning. CRISP­DM stands for Cross Industry Standard Process – Data Mining. This process involves six steps. The
steps are listed below in Figure 1.11. Understand the business Understand the data Data preprocessing Modelling
Model evaluation Model deployment Figure 1.11: A Machine Learning/Data Mining Process ML_01.indd 14 26­Mar­21
3:12:41 PM © Oxford University Press. All rights reserved. O x f o r d U n i v e r s i t y P r e s s
29. Introduction to Machine Learning 15 1. Understanding the business – This step involves understanding the
objectives and requirements of the business organization. Generally, a single data mining algorithm is enough for
giving the solution. This step also involves the formulation of the problem statement for the data mining process. 2.
Understanding the data – It involves the steps like data collection, study of the charac­ teristics of the data,
formulation of hypothesis, and matching of patterns to the selected hypothesis. 3. Preparation of data – This step
involves producing the final dataset by cleaning the raw data and preparation of data for the data mining process.
The missing values may cause problems during both training and testing phases. Missing data forces classifiers to
produce inaccurate results. This is a perennial problem for the classification models. Hence, suitable strategies
should be adopted to handle the missing data. 4. Modelling – This step plays a role in the application of data mining
algorithm for the data to obtain a model or pattern. 5. Evaluate – This step involves the evaluation of the data mining
results using statistical analysis and visualization methods. The performance of the classifier is determined by
evaluating the accuracy of the classifier. The process of classification is a fuzzy issue. For example, classification of
emails requires extensive domain knowledge and requires domain experts. Hence, performance of the classifier is
very crucial. 6. Deployment – This step involves the deployment of results of the data mining algorithm to improve
the existing process or for a new situation. 1.7 MACHINE LEARNING APPLICATIONS Machine Learning technologies
are used widely now in di몭erent domains. Machine learning appli­ cations are everywhere! One encounters many
machine learning applications in the day­to­day life. Some applications are listed below: 1. Sentiment analysis – This
is an application of natural language processing (NLP) where the words of documents are converted to sentiments
like happy, sad, and angry which are captured by emoticons e몭ectively. For movie reviews or product reviews, five
stars or one star are automatically attached using sentiment analysis programs. 2. Recommendation systems – These
are systems that make personalized purchases possible. For example, Amazon recommends users to find related
books or books bought by people who have the same taste like you, and Netflix suggests shows or related movies of
your taste. The recommendation systems are based on machine learning. 3. Voice assistants – Products like Amazon
Alexa, Microso몭 Cortana, Apple Siri, and Google Assistant are all examples of voice assistants. They take speech
commands and perform tasks. These chatbots are the result of machine learning technologies. 4. Technologies like
Google Maps and those used by Uber are all examples of machine learning which o몭er to locate and navigate
shortest paths to reduce time. The machine learning applications are enormous. The following Table 1.4 summarizes
some of the machine learning applications. ML_01.indd 15 26­Mar­21 3:12:42 PM © Oxford University Press. All rights
reserved. O x f o r d U n i v e r s i t y P r e s s
30. 16 Machine Learning Table 1.4: Applications’ Survey Table S.No. Problem Domain Applications 1. Business
Predicting the bankruptcy of a business firm 2. Banking Prediction of bank loan defaulters and detecting credit card
frauds 3. Image Processing Image search engines, object identification, image classification, and generating synthetic
images 4. Audio/Voice Chatbots like Alexa, Microso몭 Cortana. Developing chatbots for customer support, speech to
text, and text to voice 5. Telecommuni­ cation Trend analysis and identification of bogus calls, fraudulent calls and its
callers, churn analysis 6. Marketing Retail sales analysis, market basket analysis, product performance analysis,
market segmentation analysis, and study of travel patterns of customers for marketing tours 7. Games Game
programs for Chess, GO, and Atari video games 8. Natural Language Translation Google Translate, Text
summarization, and sentiment analysis 9. Web Analysis and Services Identification of access patterns, detection of e­
mail spams, viruses, personalized web services, search engines like Google, detection of promotion of user websites,
and finding loyalty of users a몭er web page layout modification 10. Medicine Prediction of diseases, given disease
symptoms as cancer or diabetes. Prediction of e몭ectiveness of the treatment using patient history and Chatbots to
interact with patients like IBM Watson uses machine learning technologies. 11. Multimedia and Security Face
recognition/identification, biometric projects like identification of a person from a large image or video database,
and applications involving multimedia retrieval 12. Scientific Domain Discovery of new galaxies, identification of
groups of houses based on house type/geographical location, identification of earthquake epicenters, and
identification of similar land use Summary 1. Machine learning can enable top management of an organization to
extract the knowledge from the data stored in various archives to facilitate decision making. 2. Machine learning is an
important subbranch of Artificial Intelligence (AI). 3. A model is an explicit description of patterns within the data. 4. A
model can be a formula, procedure or representation that can generate data decisions. 5. Humans predict by
remembering the past, then formulate the strategy and make a prediction. In the same manner, the computers can
predict by following the process. 6. Machine learning is an important branch of AI. AI is a much broader subject. The
aim of AI is to develop intelligent agents. An agent can be a robot, humans, or other autonomous systems.
ML_01.indd 16 26­Mar­21 3:12:42 PM © Oxford University Press. All rights reserved. O x f o r d U n i v e r s i t y P r e s s
31. Introduction to Machine Learning 17 7. Deep learning is a branch of machine learning. The di몭erence between
machine learning and deep learning is that models are constructed using neural network technology in deep
learning. Neural networks are models constructed based on the human neuron models. 8. Data science deals with
gathering of data for analysis. It is a broad field that includes other fields. 9. Data analytics aims to extract useful
knowledge from crude data. There are many types of analytics. Predictive data analytics is an area that is dedicated
for making predictions. Machine learning is closely related to this branch of analytics and shares almost all
algorithms. 10. One can say thus there are two types of data – labelled data and unlabelled data. The data with a
label is called labelled data and those without a label are called unlabelled data. 11. Supervised algorithms use
labelled dataset. As the name suggests, there is a supervisor or teacher component in supervised learning. A
supervisor provides the labelled data so that the model is constructed and gives test data for checking the model. 12.
Classification is a supervised learning method. The input attributes of the classification algorithms are called
independent variables. The target attribute is called label or dependent variable. The relationship between the input
and target variables is represented in the form of a structure which is called a classification model. 13. Cluster
analysis is an example of unsupervised learning. It aims to assemble objects into disjoint clusters or groups. 14.
Semi­supervised algorithms assign a pseudo­label for unlabelled data. 15. Reinforcement learning allows the agent
to interact with the environment to get rewards. The agent can be human, animal, robot, or any independent
program. The rewards enable the agent to gain experience. 16. The emerging process model for the data mining
solutions for business organizations is CRISP­DM. This model stands for Cross Industry Standard Process – Data
Mining. 17. Machine Learning technologies are used widely now in di몭erent domains. Key Terms • Machine Learning –
A branch of AI that concerns about machines to learn automatically without being explicitly programmed. • Data – A
raw fact. • Model – An explicit description of patterns in a data. • Experience – A collection of knowledge and heuristics
in humans and historical training data in case of machines. • Predictive Modelling – A technique of developing
models and making a prediction of unseen data. • Deep Learning – A branch of machine learning that deals with
constructing models using neural networks. • Data Science – A field of study that encompasses capturing of data to
its analysis covering all stages of data management. • Data Analytics – A field of study that deals with analysis of data.
• Big Data – A study of data that has characteristics of volume, variety, and velocity. • Pattern Recognition – A field of
study that analyses a pattern using machine learning algorithms. ML_01.indd 17 26­Mar­21 3:12:42 PM © Oxford
University Press. All rights reserved. O x f o r d U n i v e r s i t y P r e s s
32. 18 Machine Learning • Statistics – A branch of mathematics that deals with learning from data using statistical
methods. • Hypothesis – An initial assumption of an experiment. • Learning – Adapting to the environment that
happens because of interaction of an agent with the environment. • Label – A target attribute. • Labelled Data – A data
that is associated with a label. • Unlabelled Data – A data without labels. • Supervised Learning – A type of machine
learning that uses labelled data and learns with the help of a supervisor or teacher component. • Classification
Program – A supervisory learning method that takes an unknown input and assigns a label for it. In simple words,
finds the category of class of the input attributes. • Regression Analysis – A supervisory method that predicts the
continuous variables based on the input variables. • Unsupervised Learning – A type of machine leaning that uses
unlabelled data and groups the attributes to clusters using a trial and error approach. • Cluster Analysis – A type of
unsupervised approach that groups the objects based on attributes so that similar objects or data points form a
cluster. • Semi­supervised Learning – A type of machine learning that uses limited labelled and large unlabelled data.
It first labels unlabelled data using labelled data and combines it for learning purposes. • Reinforcement Learning – A
type of machine learning that uses agents and environment interaction for creating labelled data for learning. • Well­
posed Problem – A problem that has well­defined specifications. Otherwise, the problem is called ill­posed. •
Bias/Variance – The inability of the machine learning algorithm to predict correctly due to lack of generalization is
called bias. Variance is the error of the model for training data. This leads to problems called overfitting and
underfitting. • Model Deployment – A method of deploying machine learning algorithms to improve the existing
business processes for a new situation. Short Questions 1. Why is machine learning needed for business
organizations? 2. List out the factors that drive the popularity of machine learning. 3. What is a model? 4. Distinguish
between the terms: Data, Information, Knowledge, and Intelligence. 5. How is machine learning linked to AI, Data
Science, and Statistics? 6. List out the types of machine learning. 7. List out the di몭erences between a model and
pattern. Patterns are local and model is global for entire dataset – Justify. 8. Are classification and clustering are same
or di몭erent? Justify. ML_01.indd 18 26­Mar­21 3:12:42 PM © Oxford University Press. All rights reserved. O x f o r d U n i
versityPress
33. Introduction to Machine Learning 19 9. List out the di몭erences between labelled and unlabelled data. 10. Point
out the di몭erences between supervised and unsupervised learning. 11. What are the di몭erences between
classification and regression? 12. What is a semi­supervised learning? 13. List out the di몭erences between reinforced
learning and supervised learning. 14. List out important classification and clustering algorithms. 15. List out at least
five major applications of machine learning. Long Questions 1. Explain in detail the machine learning process model.
2. List out and briefly explain the classification algorithms. 3. List out and briefly explain the unsupervised algorithms.
Numerical Problems and Activities 1. Let us assume a regression algorithm generates a model y = 0.54 + 0.66 x for
data pertaining to week sales data of a product. Here, x is the week and y is the product sales. Find the prediction for
the 5th and 8th week. 2. Give two examples of patterns and models. 3. Survey and find out atleast five latest
applications of machine learning. 4. Survey and list out atleast five products that use machine learning. Crossword 10
15 8 9 24 21 19 18 20 23 22 13 14 16 17 11 12 7 1 2 3 4 5 6 ML_01.indd 19 26­Mar­21 3:12:42 PM © Oxford University
Press. All rights reserved. O x f o r d U n i v e r s i t y P r e s s
34. 20 Machine Learning Across 3. The initial assumption of the experiment is called a ___________. 6. A study that
deals with the analysis of data is called ___________. 11. A domain of study that covers all the aspects of data
management is called ___________ science. 13. Data is a ___________ fact. 15. CRISP­DM is a ___________ model.
17. Pattern recogition is used for identifying patterns in ___________ and videos. 19. Unsupervised learning uses
___________ data. 21. Reinforcement learning uses feedback from environment for learning ―. (True/False) 22.
Amazon Alexa is a ___________ assistant. 23. Classification is an example of ___________ learning. 24.
___________ data has the characteristics such as volume, variety and velocity. Down 1. A problem that has well­
posed specification can be solved using machine learning algorithms ―. (Yes/No) 2. Cluster analysis is an example of
___________ learning. 4. Learning from data is the aim of statistics ―. (Yes/No) 5. Predictive models can predict
based on ___________ data. 7. Regression can predict ___________ variables. 8. Learning is ___________ to the
environment. 9. Bias and variance cause overfitting and ___________ of model. 10. Lack of generalization in
machine learning happens because of Bias ―. (Yes/No) 12. Supervised learning uses ___________ data. 14. Machine
learning is ___________ learning without being explicity programmed. 16. Model is a description of ___________.
18. A semi­supervised algorithm assigns a pseudo­label for unlabelled data ―. (True/ False) 20. Machine learning
using neural networks from a domain called artificial neural network and ___________ learning. ML_01.indd 20 26­
Mar­21 3:12:42 PM © Oxford University Press. All rights reserved. O x f o r d U n i v e r s i t y P r e s s
35. Introduction to Machine Learning 21 Word Search Find and mark the words listed below. R C X L M S S H I S T O
RICALQUBNOHXBEHYPOTHESISMYESDYTYVPZUKKAHDDTHXPIUBJQTXIWTGILCRGL
DRBSUPERVISEDWPROCESSKWRFAYUNLABELLEDKSUPATTERNSRNPYNLYTDVKUXU
ZWFYESERCDXIITESDBMZLABELLEDKHWLWBYDKTISUOPNDTSWMUNDERFITTINGJC
NNPBZBPRTRAWJMVOPRPHFCWKLGFEIQUFUVQBOBWEYCLYVWXQDETPRGLXTEYJXY
OQUANTITATIVEBXVUVLHXUCPZQKWISYIVQZTKVGWIVFUTWSODRBIMAGESBOOZMA
 GDSNMZVGCNLOKYSQOMOAFITBTJQEQAZLAUTOMATICFEIJFFCMRGADKXZPVNKWM
H L O D E E P Z B B O E U C X F N D A T A D X K A N A L Y T I C S C K E K E Automatic Raw Patterns Historical Deep Data
Analytics Big Images Yes Hypothesis Adapting Labelled Supervised Quantitative Unlabelled Unsupervised True True
Yes Yes Underfitting Process Voice ML_01.indd 21 26­Mar­21 3:12:42 PM © Oxford University Press. All rights reserved.
OxfordUniversityPress

Download

AboutSupportTermsPrivacyCopyright Cookie Preferences Do not sell or share my personal informationEverand

© 2024 SlideShare from Scribd