Artificial Intelligence and Machine Learning

(CS3491 – Artificial Intelligence and Machine Learning)

As per the Latest Syllabus of Anna University, Chennai

(Regulation 2021)

Dr. S. Nithyanantham,
Assistant Professor
School of Engineering
Computer Science and Engineering
Presidency University
Bangalore.

Mr. S. Arunprasath B.E., M.E., (Ph.D.),

Research scholar,
SRM University,
Chennai

Mr. P. Krishna Sankar, B.E., M.E.,

Freelance Entrepreneur / Consultant
Kavindapadi

A.R.S. Publications
No. 11, Veerabathra Nagar, Part II,
8th Street, Medavakkam,
Chennai – 600 100, Tamil Nadu, India.
Phone: 044 – 48587467, Mobile: 9840025186
eMail: [email protected]
web: www.arspublications.com
 PREFACE
This book “Artificial Intelligence and Machine Learning” is about basic idea towards machines

are working intelligently and its designing. It provides clear view towards uniformed and heuristic

search techniques. It provides a preliminary study on reasoning under uncertainty with several

illustration. It affords procedural footsteps from artificial intelligence to machine learning.

Unit I: Introduction towards artificial intelligence and working of agents. Contributes a

knowledge on several search algorithms and constraint satisfaction problem solving

techniques with optimization.

Unit II: Outline on how machines intelligently reasoning with Bayesian based knowledge

relevance detection.

Unit III: Transitory awareness on Machine learning, regression and classification models.

Provides a procedure to perform classification and regression on data’s.

Unit IV: Brief knowledge over combining multiple learners and various ensemble techniques.

It provides detailed understanding and procedures for unsupervised learning.

Unit V: Provides a study over perceptron and its various mechanisms. Understanding

towards Unit saturation and Reinforcement learning unit.

 ACKNOWLEDGEMENT

Primarily, we would like to thank God. In the process of putting this book together, we realized

how true this gift of writing is for us to share our knowledge. You give us the power to believe in our

passion and pursue our dreams. We could never have done this without the faith we have in you, the

Almighty.

We wholeheartedly thank next God, thy Parents, for showing faith with us and giving

us liberty to choose what we desire. We salute you all for the selfless love, care, pain and sacrifice you

did to shape our life.

We sincerely thank our Colleagues, Friends and Well-wishers for their understanding, patience

in addition, constant encouragement.

Finally, we offer our special thanks to Thiru. A. Ramesh, A. R. S. Publishers and his Colleagues

for their tireless effort in overseeing the production of the book.

The authors would be happy to collect opinion for supplementary improvement of the book.

Dr. S. Nithyanantham

Mr. S. Arun Prasath

Mr. P. Krishna Sankar

 CS3491 ARTIFICIAL INTELLIGENCE AND MACHINE LEARNING LTPC

3024

UNIT I PROBLEM SOLVING 9

Introduction to AI - AI Applications - Problem solving agents – search algorithms – uninformed search

strategies – Heuristic search strategies – Local search and optimization problems – adversarial search
– constraint satisfaction problems (CSP).

UNIT II PROBABILISTIC REASONING 9

Acting under uncertainty – Bayesian inference – naïve bayes models. Probabilistic reasoning –
Bayesian networks – exact inference in BN – approximate inference in BN – causal networks.

UNIT III SUPERVISED LEARNING 9

Introduction to machine learning – Linear Regression Models: Least squares, single & multiple
variables, Bayesian linear regression, gradient descent, Linear Classification Models: Discriminant
function – Probabilistic discriminative model - Logistic regression, Probabilistic generative model –
Naive Bayes, Maximum margin classifier – Support vector machine, Decision Tree, Random forests.

UNIT IV ENSEMBLE TECHNIQUES AND UNSUPERVISED LEARNING 9

Combining multiple learners: Model combination schemes, Voting, Ensemble Learning - bagging,
boosting, stacking, Unsupervised learning: K-means, Instance Based Learning: KNN, Gaussian mixture
models and Expectation maximization.

UNIT V NEURAL NETWORKS 9

Perceptron - Multilayer perceptron, activation functions, network training – gradient descent

optimization – stochastic gradient descent, error backpropagation, from shallow networks to deep
networks –Unit saturation (aka the vanishing gradient problem) – ReLU, hyperparameter tuning, batch
normalization, regularization, dropout.
 CONTENTS
UNIT I

PROBLEM SOLVING
1.1 Introduction to AI 3

1.1.1. What is intelligence? 3

1.1.2. Definition 5

1.1.3. AI History 7

1.2 AI Applications 10

1.2.1 Applications of AI Algorithms in Video Games 10

1.2.2 Movement and Path-finding 11

1.2.3 NPC Behavior 12

1.2.4 Team AI 16

1.2.5 Real-Time Strategy AI 19

1.3 Problem solving agents 21

1.3.1 Well-defined problems and solutions 25

1.3.2 Formulating problems 27

1.4 Search algorithms 28

1.4.1 Infrastructure for search algorithms 33

1.4.2 Measuring problem-solving performance 35

1.5 Uninformed search strategies 36

1.5.1 Breadth-first search 36

1.5.2 Uniform-cost search 39

1.5.3 Depth-first search 41

1.5.4 Depth-limited search 43

1.5.5 Iterative deepening depth-first search 44

1.5.6 Bidirectional search 46

1.5.7 Comparing uninformed search strategies 48

1.6 Local Search and Optimization Problems48

1.6.1 Hill-climbing search 49

1.6.2 Simulated annealing 53

1.6.3 Local beam search 54
1.6.4 Genetic algorithms 55

1.7 Adversarial Search 58

1.7.1 Optimal Decisions in Games 62

1.7.2 Alpha - Beta Pruning 67

1.7.3 Imperfect Real-Time Decisions 73

1.8 Constraint Satisfaction Problems (CSP) 80

1.8.1 Defining Constraint Satisfaction Problems 80

1.8.2 Constraint Propagation 87

1.8.3 Backtracking Search 96

UNIT II
PROBABILISTIC REASONING

2.1 Acting under uncertainty 3

2.1.1 Summarizing uncertainty 3

2.1.2 Uncertainty and rational decisions 5

2.2 Bayesian inference 6

2.2.1 Rejection sampling in Bayesian networks 9

2.2.2 Likelihood weighting 10

2.2.3 Inference by Markov chain simulation 12

2.2.4 Gibbs sampling in Bayesian networks 13

2.2.5 Why Gibbs sampling works 13

2.3 Naive bayes models 16

2.3.1 Naive Bayes probabilistic model 16

2.3.2 Parameter estimation 18

2.3.3 Sample correction 18

2.3.4 Constructing a classifier from probability model 18

2.3.5 Sample correction 19

2.3.6 Constructing a classifier from probability model 19

2.3.7 Examples Gender Classification 19

2.3.8 Document Classification 21

2.4 Probabilistic reasoning 22

2.4.1 Representing Knowledge in an Uncertain Domain 22
2.4.2 Bayesian networks 24

2.4.3 Representing full joint distribution 24

2.4.4 A method for constructing Bayesian networks 25

2.5 Exact inference in BN 26

2.5.1 Inference by enumeration 26

2.5.2 Variable elimination algorithm 27

2.5.3 Operations on factors 29

2.5.4 Variable ordering and variable relevance 30

2.5.5 Complexity of exact inference 31

2.5.6 Clustering algorithms 32

2.6 Approximate Inference in BN 32

2.6.1 Direct sampling methods 33

2.7 Causal networks 33

2.7.1 Causal Bayesian Networks as a Visual Tool - Characterising patterns of unfairness underlying
a dataset 34
2.7.2 Causal Bayesian Networks as a Quantitative Tool - Path-specific (counterfactual) inference
techniques for fairness 36

UNIT III

SUPERVISED LEARNING

3.1. Introduction to machine learning 2

3.1.1. Examples of Machine Learning Applications 3

3.2. Linear Regression Models 7

3.2.1 Least squares 10

3.2.2 Single & multiple variables 12

3.2.3 Bayesian linear regression 15

3.2.4 Gradient descent 16

3.3. Linear Classification Models 17

3.3.1 Discriminant function 17

3.3.2 Probabilistic discriminative model 22

3.3.3 Logistic regression 23

3.3.4 Probabilistic generative model 24
3.3.5 Naive Bayes 26
3.3.6 Maximum margin classifier 28

3.4. Support vector machine29

3.4.1 Linear SVM 31

3.4.2 Non-Linear SVM 32

3.5. Decision Tree 34

3.5.1 Terminologies 35
3.5.2 Working of Algorithm 35

3.5.3 Example 36

3.5.4 Attribute Selection Measures 36

3.5.5 Information Gain 36

3.5.6 Gini Index 37

3.5.7 Pruning: Getting an Optimal Decision tree 37

3.6. Random forests 37

3.6.1 Working of algorithm 38

3.6.2 Example 39

3.6.3 Applications of Random Forest 39

UNIT IV

ENSEMBLE TECHNIQUES AND UNSUPERVISED LEARNING

4.1 Combining multiple learners 2

4.1.1 Model combination schemes 2

4.1.2 Voting 3

4.2 Ensemble Learning 6

4.2.1 Bagging 7
4.2.2 Boosting 8

4.2.3 Stacking 10

4.3 Unsupervised learning 12

4.3.1 K-means 12

4.4 Instance Based Learning16

4.4.1 KNN 17

4.4.2 Gaussian mixture models 19

4.4.3 Expectation maximization 22
UNIT V

NEURAL NETWORKS

5.1 Introduction about Neural Networks 2

5.2 Perceptron 2

5.2.1 Types of Perceptron 3

5.2.2 Characteristics of a Perceptron Model 4

5.2.3 Limitation of a Perceptron model 4

5.2.4 Perceptron Learning Rule 4

5.2.5 Perceptron Function 5

5.2.6 Inputs of a Perceptron 5

5.2.7 Activation Functions of Perceptron 6

5.2.8 Output of Perceptron 6

5.2.9 Error in Perceptron 7

5.2.10 Perceptron: Decision Function 7

5.2.11 Bias Unit 7

5.2.12 Output 7

5.3 Network Training 8

5.3.1 Gradient descent optimization 10

5.3.2 Stochastic Gradient Descent (SGD) 11

5.4 Error Backpropagation 12

5.4.1 Evaluation of error-function derivatives 13

5.4.2 Efficiency of backpropagation 14

5.4.3 From Shallow networks to deep networks 15

5.5 Unit saturation also known as(aka) the vanishing gradient problem 16

5.5.1 ReLU 18

5.5.2 Batch normalization 19

5.5.3 Hyperparameter tuning 20

5.5.4 Regularization 21

5.5.5 Dropout 23