A Seminar Report on

MACHINE LEARNING

G.B. Pant Institute of Engineering & Technology

Department of Electronics and Communication Engineering

<2017-2018>

Guided By: - Submitted By: -

Prof. Dr. M.K . Aggarwal Harshit Semwal
Asst. Professor III YEAR (VI SEM)
 ACKNOWLEDGEMENT

It is a matter of great pleasure for me to submit this report on MACHINE LEARNING, as a part
of curriculum for award of BACHELOR’S IN TECHNOLOGY (ECE) degree of Uttrakhand
Technical University, Dehradun (Uttrakhand).

At this moment of accomplishment, I am presenting my work with great pride and pleasure, I
would like to express my sincere gratitude to all those who helped me in the successful
completion of my venture. I would like to thank our PROF.M.K. AGGARWAL for helping me
in the successful accomplishment of my study and for her timely and valuable suggestions. His
constructive criticism has contributed immensely to the evolution of my ideas on the subject.

I am exceedingly grateful to my Head of Department PROF. Y SINGH and other faculty

members for their inspiration and encouragement. I would also like to thank my parents and
friends for their over whelming and whole hearted encouragement and support without which
this would not have been successful.

HARSHIT SEMWAL

2
G.B PANT INSTITUTE OF ENGINEERING & TECHNOLOGY ,GHURDAURI(PAURI)
DEPARTMENT OF ELECTRONICS AND COMMUNICATION ENGINEERING

CERTIFICATE

This is to certify that the report entitled “MACHINE LEARNING” has been carried out
by HARSHIT SEMWAL under my guidance in partial fulfillment of the degree of Bachelor of
Technology in ELECTRONICS AND COMMUNICATION ENGINEERING of Uttrakhand
Technical University, Ghurdauri(Pauri) during the academic year 2017-2018.

(Prof. M.K Aggarwal)

Lecturer Examiner

(Prof. Y Singh)

Head of Department

ECE

3
 ABSTRACT

Present day computer applications require the representation of huge amount of complex
knowledge and data in programs and thus require tremendous amount of work. Our ability to
code the computers falls short of the demand for applications. If the computers are endowed with
the learning ability, then our burden of coding the machine is eased (or at least reduced). This is
particularly true for developing expert systems where the "bottle-neck" is to extract the expert’s
knowledge and feed the knowledge to computers. The present day computer programs in general
(with the exception of some Machine Learning programs) cannot correct their own errors or
improve from past mistakes, or learn to perform a new task by analogy to a previously seen task.
In contrast, human beings are capable of all the above. Machine Learning will produce smarter
computers capable of all the above intelligent behavior.

The area of Machine Learning deals with the design of programs that can learn rules from
data, adapt to changes, and improve performance with experience. In addition to being one of the
initial dreams of Computer Science, Machine Learning has become crucial as computers are
expected to solve increasingly complex problems and become more integrated into our daily
lives. This is a hard problem, since making a machine learn from its computational tasks requires
work at several levels, and complexities and ambiguities arise at each of those levels.

So, here we study how the Machine learning take place, what are the methods, remedies
associated, applications, present and future status of machine learning.

4
 Index
ACKNOWLEDGEMENT
CERTIFICATE
ABSTRACT
Chapter 1 Introduction to Machine Learning 6

1.1 WHY MACHINE LEARNING?

Chapter 2 Learning means? 9

2.1 THE ARCHITECTURE OF A LEARNING AGENT

Chapter 3 History of Machine leaning 12

3.1 The Neural Modeling (Self Organized System)

3.2 The Symbolic Concept Acquisition Paradigm
3.3 The Modern Knowledge-Intensive Paradigm

Chapter 4 Wellsprings of Machine Learning 14

4.1 Statistics
4.2 Brain Models
4.3 Adaptive Control Theory
4.4 Psychological Models
4.5 Artificial Intelligence
4.6 Evolutionary Models

Chapter 5 Machine Learning Overview 16

5.1 The Aim of Machine Learning

5.2 Machine Learning as a Science

Chapter 6 Classification of Machine Learning 18

Chapter 7 Types of Machine Learning Algorithms 21

7.1 Algorithm Types

7.2 Machine Learning Applications
7.3 Examples of Machine Learning Problems

Chapter 8 Future Directions 28

8.1 Conclusions

REFERENCES 30

5
 Chapter 1

Introduction to Machine Learning

Machine Learning (ML) is the computerized approach to analyzing computational work

that is based on both a set of theories and a set of technologies. And, being a very active area of
research and development, there is not a single agreed-upon definition that would satisfy
everyone, but there are some aspects, which would be part of any knowledgeable person’s
definition. The definition mostly offers is:

Definition: Ability of a machine to improve its own performance through the use of
a software that employs artificial intelligence techniques to mimic the ways by which humans
seem to learn, such as repetition and experience.

Machine Learning (ML) is a sub-field of Artificial Intelligence (AI) which concerns with
developing computational theories of learning and building learning machines. The goal of
machine learning, closely coupled with the goal of AI, is to achieve a thorough understanding
about the nature of learning process (both human learning and other forms of learning), about the
computational aspects of learning behaviors, and to implant the learning capability in computer
systems. Machine learning has been recognized as central to the success of Artificial
Intelligence, and it has applications in various areas of science, engineering and society.

1.1 WHY MACHINE LEARNING?

To answer this question, we should look at two issues:

(1).What are the goals of machine learning;

(2). Why these goals are important and desirable.

6
1.1.1 The Goals of Machine Learning.

The goal of ML, in simples words, is to understand the nature of (human and other forms of)
learning, and to build learning capability in computers. To be more specific, there are three
aspects of the goals of ML.

(1) To make the computers smarter, more intelligent. The more direct objective in this
aspect is to develop systems (programs) for specific practical learning tasks in application
domains.

(2) To develop computational models of human learning process and perform computer
simulations. The study in this aspect is also called cognitive modeling.

(3) To explore new learning methods and develop general learning algorithms
independent of applications.

1.1.2 Why the goals of ML are important and desirable.

It is self-evident that the goals of ML are important and desirable. However, we still give
some more supporting argument to this issue.

First of all, implanting learning ability in computers is practically necessary. Present day
computer applications require the representation of huge amount of complex knowledge and data
in programs and thus require tremendous amount of work. Our ability to code the computers falls
short of the demand for applications. If the computers are endowed with the learning ability, then
our burden of coding the machine is eased (or at least reduced). This is particularly true for
developing expert systems where the "bottle-neck" is to extract the expert’s knowledge and feed
the knowledge to computers. The present day computer programs in general (with the exception
of some ML programs) cannot correct their own errors or improve from past mistakes, or learn to
perform a new task by analogy to a previously seen task. In contrast, human beings are capable
of all the above. ML will produce smarter computers capable of all the above intelligent
behavior.

Second, the understanding of human learning and its computational aspect is a worthy
scientific goal. We human beings have long been fascinated by our capabilities of intelligent

7
behaviors and have been trying to understand the nature of intelligence. It is clear that central to
our intelligence is our ability to learn. Thus a thorough understanding of human learning process
is crucial to understand human intelligence. ML will gain us the insight into the underlying
principles of human learning and that may lead to the discovery of more effective education
techniques. It will also contribute to the design of machine learning systems.

Finally, it is desirable to explore alternative learning mechanisms in the space of all

possible learning methods. There is no reason to believe that the way human being learns is the
only possible mechanism of learning. It is worth exploring other methods of learning which may
be more efficient, effective than human learning.

We remark that Machine Learning has become feasible in many important applications
(and hence the popularity of the field) partly because the recent progress in learning algorithms
and theory, the rapidly increase of computational power, the great availability of huge amount of
data, and interests in commercial ML application development.

Moreover we note that ML is inherently a multi-disciplinary subject area.

We compare the human learning with machine learning along the dimensions of speed, ability to
transfer, and others. which shows that machine learning is both an opportunity and challenge, in
the sense that we can hope to discover ways for machine to learn which are better than ways
human learn (the opportunity), and that there are amply amount of difficulties to be overcome in
order to make machines learn (the challenge).

8
 Chapter 2

Learning means?

Learning is a phenomenon and process which has manifestations of various aspects.

Roughly speaking, learning process includes (one or more of) the following:

(1) Acquisition of new (symbolic) knowledge. For example, learning mathematics is this
kind of learning. When we say someone has learned math, we mean that the learner obtained
descriptions of the mathematical concepts, understood their meaning and their relationship with
each other. The effect of learning is that the learner has acquired knowledge of mathematical
systems and their properties, and that the learner can use this knowledge to solve math problems.
Thus this kind of learning is characterized as obtaining new symbolic information plus the ability
to apply that information effectively.

(2) Development of motor or cognitive skills through instruction and practice. Examples
of this kind of learning are learning to ride a bicycle, to swim, to play piano, etc. This kind of
learning is also called skill refinement. In this case, just acquiring a symbolic description of the
rules to perform the task is not sufficient, repeated practice is needed for the learner to obtain the
skill. Skill refinement takes place at the subconscious level.

(3) Refinement and organization of knowledge into more effective representations or

more useful form. One example of this kind of learning can be reorganization of the rules in a
knowledge base such that more important rules are given higher priorities so that they can be
used more easily and conveniently.

(4) Discovery of new facts and theories through observation and experiment. For
example, the discovery of physics and chemistry laws.

The general effect of learning in a system is the improvement of the system’s capability
to solve problems. It is hard to imagine a system capable of learning cannot improve its problem-
solving performance. A system with learning capability should be able to do self-changing in
order to perform better in its future problem-solving.

9
 We also note that learning cannot take place in isolation: We typically learn something
(knowledge K) to perform some tasks (T), through some experience E, and whether we have
learned well or not will be judged by some performance criteria P at the task T. For example, as
Tom Mitchell put it in his ML book, for the "checkers learning problem", the task T is to play the
game of checkers, the performance criteria P could be the percentage of games won against
opponents, and the experience E could be in the form playing practice games with a teacher (or
self). For learning to take place, we do need a learning algorithm A for self-changing, which
allows the learner to get experience E in the task T, and acquire knowledge K (thus change the
learner’s knowledge set) to improve the learner’s performance at task T.

Learning = Improving performance P at task T by

acquiring knowledge K using self-changing algorithm
A through experience E in an environment for
task T.

There are various forms of improvement of a system’s problem-solving ability:

(1) To solve wider range of problems than before - perform generalization.

(2) To solve the same problem more effectively - give better quality solutions.

(3) To solve the same problem more efficiently - faster.

There are other view points as to what constitutes the notion of learning. For example,

Minsky gives a more general definition,

"Learning is making useful changes in our minds".

McCarthy suggests,

"Learning is constructing or modifying representations of what is being experienced."

Simon suggests,

10
“Learning denotes changes in the system that are adaptive in the sense that they enable the
system to do the same task or tasks drawn from the same population more effectively the next
time”.

From this perspective, the central aspect of learning is acquisition of certain forms of
representation of some reality, rather than the improvement of performance. However, since it is
in general much easier to observe a system’s performance behavior than its internal
representation of reality, we usually link the learning behavior with the improvement of the the
system’s performance.

2.1 THE ARCHITECTURE OF A LEARNING AGENT

11
 Chapter 3

History of Machine leaning

Over the years, research in machine learning has been pursued with varying degrees of
intensity, using different approaches and placing emphasis on different, aspects and goals.
Within the relatively short history of this discipline, one may distinguish three major periods,
each centered on a different concept:

• neural modeling and decision-theoretic techniques

• symbolic concept-oriented learning
• knowledge-intensive approaches combining various learning strategies

3.1 The Neural Modeling (Self Organized System)

The distinguishing feature of the first concept was the interest in building general purpose
learning systems that start with little or no initial structure or task-oriented knowledge. The
major thrust of research based on this approach involved constructing a variety of neural model-
based machines, with random or partially random initial structure. These systems were generally
referred to as neural networks or self-organizing systems. Learning in such systems consisted of
incremental changes in the probabilities that neuron-like elements would transmit a signal. Due
to the early computer technology, most of the research under this neural network model was
either theoretical or involved the construction of special purpose experimental hardware systems.
Related research involved the simulation of evolutionary processes that through random
mutation and “natural” selection might create a system capable of some intelligent, behavior.
Experience in the above areas spawned the new discipline of pattern recognition and led to the
development of a decision-theoretic approach to machine learning. In this approach, learning is
equated with the acquisition of linear, polynomial, or related discriminated functions from a
given set of training examples. One of the best known successful learning systems utilizing such
techniques as well as some original new ideas involving non-linear transformations was

12
Samuel’s checkers program. Through repeated training, this program acquired master-level
performance somewhat; different, but closely related, techniques utilized methods of statistical
decision theory for learning pattern recognition rules.

3.2 The Symbolic Concept Acquisition Paradigm

A second major paradigm started to emerge in the early sixties stemming from the work
of psychologist and early AI researchers on models of human learning by Hunt. The paradigm
utilized logic or graph structure representations rather than numerical or statistical methods
Systems learned symbolic descriptions representing higher level knowledge and made strong
structural assumptions about the concepts to be acquired. Examples of work in this paradigm
include research on human concept acquisition and various applied pattern recognition systems.

3.3 The Modern Knowledge-Intensive Paradigm

The third paradigm represented the most recent period of research starting in the mid
seventies. Researchers have broadened their interest beyond learning isolated concepts from
examples, and have begun investigating a wide spectrum of learning methods, most based upon
knowledge-rich systems specifically, this paradigm can be characterizing by several new trends,
including:

1. Knowledge-Intensive Approaches: Researchers are strongly emphasizing the use of task-

oriented knowledge and the constraints it provides in guiding the learning process One lesson
from the failures of earlier knowledge and poor learning systems that is acquire and to acquire
new knowledge a system must already possess a great deal of initial knowledge

2. Exploration of alternative methods of learning: In addition to the earlier research emphasis

on learning from examples, researchers are now investigating a wider variety of learning
methods such as learning from instruction.

In contrast to previous efforts, a number of current systems are incorporating abilities to generate
and select tasks and also incorporate heuristics to control their focus of attention by generating
learning tasks, proposing experiments to gather training data, and choosing concepts to acquire

13
 Chapter 4

Wellsprings of Machine Learning

Work in machine learning is now converging from several sources. These different
traditions each bring different methods and different vocabulary which are now being assimilated
into a more unified discipline. Here is a brief listing of some of the separate disciplines that have
contributed to machine learning;

4.1 Statistics

A long-standing problem in statistics is how best to use samples drawn from unknown
probability distributions to help decide from which distribution some new sample is drawn. A
related problem is how to estimate the value of an unknown function at a new point given the
values of this function at a set of sample points. Statistical methods for dealing with these
problems can be considered instances of machine learning because the decision and estimation
rules depend on a corpus of samples drawn from the problem environment.

4.2 Brain Models

Non-linear elements with weighted inputs have been suggested as simple models of
biological neurons. Brain modelers are interested in how closely these networks approximate the
learning phenomena of living brains. Several important machine learning techniques are based
on networks of nonlinear elements often called neural networks. Work inspired by this school is
sometimes called connectionism, brain-style computation, or sub-symbolic processing.

4.3 Adaptive Control Theory

Control theorists study the problem of controlling a process having unknown parameters
which must be estimated during operation. Often, the parameters change during operation, and

14
the control process must track these changes. Some aspects of controlling a robot based on
sensory inputs represent instances of this sort of problem.

4.4 Psychological Models

Psychologists have studied the performance of humans in various learning tasks. An early
example is the EPAM network for storing and retrieving one member of a pair of words when
given another. Related work led to a number of early decision tree and semantic network
methods. More recent work of this sort has been influenced by activities in artificial.

4.5 Artificial Intelligence

From the beginning, AI research has been concerned with machine learning. Samuel
developed a prominent early program that learned parameters of a function for evaluating board
positions in the game of checkers. AI researchers have also explored the role of analogies in
learning and how future actions and decisions can be based on previous exemplary cases. Recent
work has been directed at discovering rules for expert systems using decision-tree methods and
inductive logic programming.
Another theme has been saving and generalizing the results of problem solving using
explanation-based learning.

4.6 Evolutionary Models

In nature, not only do individual animals learn to perform better, but species evolve to be
better but in their individual niches. Since the distinction between evolving and learning can be
blurred in computer systems, techniques that model certain aspects of biological evolution have
been proposed as learning methods to improve the performance of computer programs. Genetic
algorithms and genetic programming are the most prominent computational techniques for
evolution.

15
 Chapter 5

Machine Learning Overview

Machine Learning can still be defined as learning the theory automatically from the data,
through a process of inference, model fitting, or learning from examples:

 Automated extraction of useful information from a body of data by building good

probabilistic models.

 Ideally suited for areas with lots of data in the absence of a general theory.

5.1 The Aim of Machine Learning

The field of machine learning can be organized around three primary research Areas:

 Task-Oriented Studies: The development and analysis of learning systems oriented

toward solving a predetermined set, of tasks (also known as the “engineering approach”).

 Cognitive Simulation: The investigation and computer simulation of human learning

processes (also known as the “cognitive modeling approach”)

 Theoretical Analysis: The theoretical exploration of the space of possible learning

methods and algorithms independent application domain.

Although many research efforts strive primarily towards one of these objectives, progress in
on objective often lends to progress in another. For example, in order to investigate the space of
possible learning methods, a reasonable starting point may be to consider the only known
example of robust learning behavior, namely humans (and perhaps other biological systems)
Similarly, psychological investigations of human learning may held by theoretical analysis that
may suggest various possible learning models. The need to acquire a particular form of

16
knowledge in stone task-oriented study may itself spawn new theoretical analysis or pose the
question: “how do humans acquire this specific skill (or knowledge)?” The existence of these
mutually supportive objectives reflects the entire field of artificial intelligence where expert
system research, cognitive simulation, and theoretical studies provide some cross-fertilization of
problems and ideas.

5.2 Machine Learning as a Science

The clear contender for a cognitive invariant in human is the learning mechanism which
is the ability facts, skills and more abstractive concepts. Therefore understanding human learning
well enough to reproduce aspect of that learning behavior in a computer system is, in itself, a
worthy scientific goal. Moreover, the computer can render substantial assistance to cognitive
psychology, in that it may be used to test the consistency and completeness of learning theories
and enforce a commitment to the fine-structure process level detail that precludes meaningless
tautological or untestable theories (Bishop, 2006).
The study of human learning processes is also of considerable practical significance. Gaining
insights into the principles underlying human learning abilities is likely to lead to more effective
educational techniques. Machine learning research is all about developing intelligent computer
assistant or a computer tutoring systems and many of these goals are shared within the machine
learning fields. According to Jaime et al who stated computer tutoring are starting to incorporate
abilities to infer models of student competence from observed performance. Inferring the scope
of a student’s knowledge and skills in a particular area allows much more effective and
individualized tutoring of the student.

17
 Chapter 6

Classification of Machine Learning

There are several areas of machine learning that could be exploited to solve the problems
of email management and our approach implemented unsupervised machine learning method.
Unsupervised learning is a method of machine learning whereby the algorithm is presented
with examples from the input space only and a model is fit to these observations. In unsupervised
learning, a data set of input objects is gathered. Unsupervised learning then typically treats input
objects as a set of random variables. A joint density model is then built for the data set. The
problem of unsupervised learning involved learning patterns in the input when no specific
output values are supplied”.
In the unsupervised learning problem, we observe only the features and have no measurements
of the outcome. Our task is rather to describe how the data are organized or clustered”. Trevor
Hastie explained that "In unsupervised learning or clustering there is no explicit teacher, and the
system forms clusters or ‘natural groupings’ of the input patterns. “Natural” is always defined
explicitly or implicitly in the clustering system itself; and given a particular set of patterns or cost
function; different clustering algorithms lead to different clusters. Often the user will set the
hypothesized number of different clusters ahead of time, but how should this be done?
According to Richard O. Duda, “How do we avoid inappropriate representations?"
There are various categories in the field of artificial intelligence. The classifications of machine
learning systems are:

 Supervised Machine Learning: Supervised learning is a machine learning technique for

learning a function from training data. The training data consist of pairs of input objects
(typically vectors), and desired outputs. The output of the function can be a continuous
value (called regression), or can predict a class label of the input object (called
classification).

18
The task of the supervised learner is to predict the value of the function for any valid
input object after having seen a number of training examples (i.e. pairs of input and target
output). To achieve this, the learner has to generalize from the presented data to unseen
situations in a "reasonable" way. Supervised learning is a machine learning technique
whereby the algorithm is first presented with training data which consists of examples
which include both the inputs and the desired outputs; thus enabling it to learn a function.
The learner should then be able to generalize from the presented data to unseen
examples." by Mitchell. Supervised learning also implies we are given a training set of
(X, Y) pairs by a “teacher”. We know (sometimes only approximately) the values of f

19
 for the m samples in the training set, ≡ we assume that if we can find a hypothesis, h,
that closely agrees with f for the members of ≡ then this hypothesis will be a good guess
for f especially if ≡ is large. Curve fitting is a simple example of supervised learning of a
function.

• Unsupervised Machine Learning: Unsupervised learning is a type of machine learning

where manual labels of inputs are not used. It is distinguished from supervised learning
approaches which learn how to perform a task, such as classification or regression, using
a set of human prepared examples. Unsupervised learning means we are only given the
Xs and some (ultimate) feedback function on our performance. We simply have a
training set of vectors without function values of them. The problem in this case,
typically, is to partition the training set into subsets, ≡1 ……≡ R , in some appropriate
way.

20
 Chapter 7

Types of Machine Learning Algorithms

Machine learning algorithms are organized into taxonomy, based on the desired outcome of
the algorithm. Common algorithm types include:

 Supervised learning → where the algorithm generates a function that maps inputs to
desired outputs. One standard formulation of the supervised learning task is the
classification problem: the learner is required to learn (to approximate the behavior of) a
function which maps a vector into one of several classes by looking at several input-
output examples of the function.

 Unsupervised learning →which models a set of inputs, labeled examples are not
available.

 Semi-supervised learning → which combines both labeled and unlabeled examples to

generate an appropriate function or classifier.

 Reinforcement learning → where the algorithm learns a policy of how to act given an
observation of the world. Every action has some impact in the environment, and the
environment provides feedback that guides the learning algorithm.

 Transduction → similar to supervised learning, but does not explicitly construct a

function: instead, tries to predict new outputs based on training inputs, training outputs,
and new inputs.

 Learning to learn → where the algorithm learns its own inductive bias based on previous
experience.

The performance and computational analysis of machine learning algorithms is a branch of

statistics known as computational learning theory. Machine learning is about designing

21
algorithms that allow a computer to learn. Learning is not necessarily involves consciousness but
learning is a matter of finding statistical regularities or other patterns in the data. Thus, many
machine learning algorithms will barely resemble how human might approach a learning task.
However, learning algorithms can give insight into the relative difficulty of learning in different
environments.

7.1 Algorithm Types

In the area of supervised learning which deals much with classification. These are the
algorithms types:

a. Linear Classifiers

1. Fisher’s linear discriminant

2. Naïve Bayes Classifier

3. Perceptron

4. Support Vector Machine

b. Quadratic Classifiers

c. Boosting

d. Decision Tree

e. Neural networks

f. Bayesian Networks

7.1. a Linear Classifiers:

In machine learning, the goal of classification is to group items that have similar feature
values, into groups. Timothy et al (Timothy Jason Shepard, 1998) stated that a linear classifier
achieves this by making a classification decision based on the value of the linear combination of
the features. If the input feature vector to the classifier is a real vector x, then the output score is

22
where is a real vector of weights and f is a function that converts the dot product of the two
vectors into the desired output.

7.1. (a.1) Fisher’s linear discriminant

Linear discriminant analysis (LDA) and the related Fisher's linear discriminant are
methods used in machine learning to find a linear combination of features which characterizes
or separates two or more classes of objects or events. The resulting combination may be used as
a linear classifier or, more commonly, for dimensionality reduction before later classification.

7.1. (a.2) Naïve Bayes Classifier

A naive Bayes classifier is a simple probabilistic classifier based on applying Bayes'

theorem with strong (naive) independence assumptions. A more descriptive term for the
underlying probability model would be "independent feature model".
In simple terms, a naive Bayes classifier assumes that the presence or absence of a particular
feature is unrelated to the presence or absence of any other feature, given the class variable. For
example, a fruit may be considered to be an apple if it is red, round, and about 3" in diameter. A
naive Bayes classifier considers each of these features to contribute independently to the
probability that this fruit is an apple, regardless of the presence or absence of the other features.

7.1. (a.3) Perceptron

The perceptron is an algorithm for supervised classification of an input into one of

several possible non-binary outputs. The learning algorithm for perceptrons is an online
algorithm, in that it processes elements in the training set one at a time.

7.1. (a.4) Support vector machines

In machine learning, support vector machines (SVMs) are supervised learning models
with associated learning algorithms that analyze data and recognize patterns, used for

23
classification and regression analysis. The basic SVM takes a set of input data and predicts, for
each given input, which of two possible classes forms the output, making it a non-
probabilistic binary linear classifier. Given a set of training examples, each marked as belonging
to one of two categories, an SVM training algorithm builds a model that assigns new examples
into one category or the other. An SVM model is a representation of the examples as points in
space, mapped so that the examples of the separate categories are divided by a clear gap that is as
wide as possible. New examples are then mapped into that same space and predicted to belong to
a category based on which side of the gap they fall on.

7.1.b Quadratic classifier

A quadratic classifier is used in machine learning and statistical classification to separate

measurements of two or more classes of objects or events by a quadric surface. It is a more
general version of the linear classifier.

7.1.c Boosting

Boosting is a machine learning meta-algorithm for reducing bias in supervised learning.

Boosting is based on the question posed as “Can a set of weak learners create a single strong
learner?” A weak learner is defined to be a classifier which is only slightly correlated with the
true classification. In contrast, a strong learner is a classifier that is arbitrarily well-correlated
with the true classification.

7.1.d Neural networks

Neural networks are capable of machine learning and pattern recognition. They are
usually presented as systems of interconnected "neurons" that can compute values from inputs by
feeding information through the network. Neural networking is the science of creating
computational solutions modeled after the brain. Like the human brain, neural networks are
trainable-once they are taught to solve one complex problem, they can apply their skills to a new
set of problems without having to start the learning process from scratch.

7.1.e Bayesian network

24
 A Bayesian network, Bayes network, belief network, Bayes(ian) model or probabilistic
directed acyclic graphical model is a probabilistic graphical model (a type ofstatistical model)
that represents a set of random variables and their conditional dependencies via a directed acyclic
graph (DAG). For example, suppose that there are two events which could cause grass to be wet:
either the sprinkler is on or it's raining. Also, suppose
that the rain has a direct effect on the use of the
sprinkler (namely that when it rains, the sprinkler is
usually not turned on). Then the situation can be
modeled with a Bayesian network (shown). All three
variables have two possible values, T (for true) and F
(for false).

7.1.f Decision Trees

A decision tree is a hierarchical data structure implementing the divide-and-conquer

strategy. It is an efficient nonparametric method, which can be used for both classification and
regression. A decision tree is a hierarchical model for supervised learning whereby the local
region is identified in a sequence of recursive splits in a smaller number of steps. A decision tree
is composed of internal decision nodes and terminal leaves (see figure). Each decision node m
implements a test function fm(x) with discrete outcomes labeling the branches. Given an input, at
each node, a test is applied and one of the branches is taken depending on the outcome. This
process starts at the root and is repeated recursively until a leaf node is hit, at which point the
value written in the leaf constitutes the output.

25
7.2 Machine Learning Applications

The other aspect for classifying learning systems is the area of application which gives a
new dimension for machine learning. Below are areas to which various existing learning systems
have been applied. They are:

1) Computer Programming

2) Game playing (chess, poker, and so on)

3) Image recognition, Speech recognition

4) Medical diagnosis

5) Agriculture, Physics

6) Email management, Robotics

7) Music

8) Mathematics

26
9) Natural Language Processing and many more.

7.3 Examples of Machine Learning Problems

There are many examples of machine learning problems. Much of this course will focus
on classification problems in which the goal is to categorize objects into a fixed set of categories.
Here are several examples:

• Optical character recognition: categorize images of handwritten characters by the letters

represented

• Face detection: find faces in images (or indicate if a face is present)

• Spam filtering: identify email messages as spam or non-spam

• Topic spotting: categorize news articles (say) as to whether they are about politics, sports,
entertainment, etc.

• Spoken language understanding: within the context of a limited domain, determine the meaning
of something uttered by a speaker to the extent that it can be classified into one of a fixed set of
categories

• Medical diagnosis: diagnose a patient as a sufferer or non-sufferer of some disease

• Customer segmentation: predict, for instance, which customers will respond to a particular
promotion

• Fraud detection: identify credit card transactions (for instance) which may be fraudulent in
nature

• Weather prediction: predict, for instance, whether or not it will rain tomorrow

27
 Chapter 8

Future Directions

Research in Machine Learning Theory is a combination of attacking established

fundamental questions, and developing new frameworks for modeling the needs of new machine
learning applications. While it is impossible to know where the next breakthroughs will come, a
few topics one can expect the future to hold include:

• Better understanding how auxiliary information, such as unlabeled data, hints from a
user, or previously-learned tasks, can best be used by a machine learning algorithm to improve
its ability to learn new things. Traditionally, Machine Learning Theory has focused on problems
of learning a task (say, identifying spam) from labeled examples (email labeled as spam or not).
However, often there is additional information available. One might have access to large
quantities of unlabeled data (email messages not labeled by their type, or discussion-group
transcripts on the web) that could potentially provide useful information. One might have other
hints from the user besides just labels, e.g. highlighting relevant portions of the email message.
Or, one might have previously learned similar tasks and want to transfer some of that experience
to the job at hand. These are all issues for which a solid theory is only beginning to be
developed.

• Further developing connections to economic theory. As software agents based on

machine learning are used in competitive settings, “strategic” issues become increasingly
important. Most algorithms and models to date have focused on the case of a single learning
algorithm operating in an environment that, while it may be changing, does not have its own
motivations and strategies. However, if learning algorithms are to operate in settings dominated
by other adaptive algorithms acting in their own users’ interests, such as bidding on items or
performing various kinds of negotiations, then we have a true merging of computer science and
economic models. In this combination, many of the fundamental issues are still wide open.

28
• Development of learning algorithms with an eye towards the use of learning as part of a
larger system. Most machine learning models view learning as a standalone process, focusing on
prediction accuracy as the measure of performance. However, when a learning algorithm is
placed in a larger system, other issues may come into play. For example, one would like
algorithms that have more powerful models of their own confidence or that can optimize
multiple objectives. One would like models that capture the process of deciding what to learn, in
addition to how to learn it. There has been some theoretical work on these issues, but there is
certainly is much more to be done.

8.1 Conclusions

Machine Learning Theory is both a fundamental theory with many basic and compelling
foundational questions, and a topic of practical importance that helps to advance the state of the
art in software by providing mathematical frameworks for designing new machine learning
algorithms. It is an exciting time for the field, as connections to many other areas are being
discovered and explored, and as new machine learning applications bring new questions to be
modeled and studied. It is safe to say that the potential of Machine Learning and its theory lie
beyond the frontiers of our imagination.

29
 REFERENCES

• Alpaydin, E. (2004). Introduction to Machine Learning. Massachusetts, USA: MIT Press.

• http://www.intechopen.com/books/new-advances-in-machine-learning/types-of-machine-
learning-algorithms

• Carling, A. (1992). Introducing Neural Networks. Wilmslow, UK: Sigma Press

• Friedberg, R. M. (1958). A learning machine: Part, 1. IBM Journal, 2-13.

• Mitchell, T. M. (2006). The Discipline of Machine Learning. Machine Learning

Department technical report CMU-ML-06-108, Carnegie Mellon University.

• Richard S. Sutton, A. G. (1998). Reinforcement Learning. MIT Press.

• Ripley, B. (1996). Pattern Recognition and Neural Networks. Cambridge University

Press.

• Tom, M. (1997). Machine Learning. Machine Learning, Tom Mitchell, McGraw Hill,
1997: McGraw Hill.

• Nilsson, N. J. (1965) Learning machines New York: McGraw-Hill Popper.

• Rosenblatt, F. (1958) “The perceptron: a probabilistic model for information storage and
organization in the brain” Psychological Review.

• Michalski, R S , Carhoncll, .J G , & Mitchcll, T. M. (1983) (Eds) Machine Learning, an

Artificial Intelligence Approach Palo Alto, CA: Tioga Press.

• Anderson, J. A. (1983) Acquisition of proof skills in geometry.In R. S. Michalski, J. G.

Carbonell & T M Mitchell (Eds ), Machine Learnzng, An Artzficaal Intelligence
Approach Palo Alto,

30