See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/329609597

Machine Learning: A Review of Techniques of Machine Learning

Article · December 2018

CITATIONS READS

0 924

3 authors, including:

Aakanksha Sharma Ankush Sharma

Federation University Australia Chandigarh University
11 PUBLICATIONS   7 CITATIONS    12 PUBLICATIONS   8 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Minimization of handover View project

All content following this page was uploaded by Ankush Sharma on 13 December 2018.

The user has requested enhancement of the downloaded file.

JASC: Journal of Applied Science and Computations ISSN NO: 1076-5131

Machine Learning: A Review of Techniques of Machine

Learning
Pooja, Student, Aakanksha Sharma, Ankush Sharma, Assistant
Department of CSE, Assistant Professor, Professor, Department of
Chandigarh University, Department of CSE, CSE, Chandigarh
Gharuan, Punjab, India Chandigarh University, University, Gharuan,
Gharuan, Punjab, India Punjab, India

Abstract: Nowadays everywhere we need process to generate the algorithms on the basis of
machines and automation to process our daily the previous inputs and past experiences.
tasks. Machine learning is one of those methods
that we use to simplify our day-to-day problems. Paper outline: First the meaning of machine
Machine learning is basically a field of computer learning is explained with some examples in sect.
science and a core branch of AI which uses [2].After that sect. [3] takes a closer look at why
statistics to give outputs. Generally machine machine learning is required in daily life. Then the
learning is the ability of a particular machine to techniques are explained briefly in sect. [4]. In sect.
learn from its previous outputs and algorithms, so [5], the conclusion and the references are given of
that it can improve on its own and don’t need our the machine learning. Now the question arises:
regular guidance to update its system. Machine What is machine learning? Let’s answer this
learning defines the steps to supervise machine’s question.
performance by learning from its historical
inputs. It is the application of artificial a) Learn from experience: This means that
intelligence and used to describe the algorithms machine can learn from its past experience
based on historical data. Nowadays some and able to generate the algorithms on
statistical experts announced that they will adopt their own to improve its new outputs.
the machine learning algorithms and name that b) Follow instructions: This means that
field as Statistical Learning. Machine learning machine should follow its past and
focuses on the development of programs in previous instructions to generate its new
computer system that can access data and use it to and better performance.
learn for itself.
Now I’ll explain you with some examples that what
Keywords: Machine Learning, Artificial is machine learning and how statistics is used in
Intelligence. machine learning.

I. INTRODUCTION Example 1: Let us suppose a computer program A

learn from its previous inputs I to compute some
The main purpose of this review paper is to tasks T to generate the improved performance P.
highlight the applications and tasks of machine
learning. This review paper will help you to gain
some useful knowledge about machine learning Previous input A Performance

and it’ll make you excited to know and search more computes
about this wonderful language-machine Learning.
It is related to computational statistics which I P
focuses on making predictions by the machines T
only. Machine learning also uses mathematics in
order to achieve theorems and computational
methods to generate its applications.

The name Machine learning was given by

Arthur Samuel in 1959.Machine learning is the Fig.1 showing example of machine learning

Volume 5, Issue 7, July /2018. Page No:538

JASC: Journal of Applied Science and Computations ISSN NO: 1076-5131

In the above example, program will generate the

output that is the algorithm after computing the
tasks on the basis of its historic experience and then
generates the performance that is better than before
and this process of constructing algorithms is called
as machine learning.

The examples provided will explain you the

definition of machine learning in a better way.

Here we have the explanation given by pictures:

Fig.4 showing performance graph of a

machine
II. WHY MACHINE LEARNING?

To understand the use of machine learning, we’ll

consider some of the examples where the concept
machine learning is applied.

Fig.2 shows the training given to machine  Such examples are: self-driving car,
Google listening search, dashboards on
In the above picture shown, the programmer is Amazon, showcases on Netflix and many
training the machine with the information he have more are the examples of machine
with him and which is important according to him learning.
and then machine will generate the output. Here  Now just imagine that you are using a
machine is learning by the programmer and giving news application but the app is unable to
the output on the basis of the given input. provide you with latest notifications and
updates on time.
 This will make you think that the app is of
no use and it is not making any aid to me.
So in this case the concept of machine
learning is used through which we can
achieve the best from the application or
software.
 Concept of machine learning enables the
feature of providing notifications
automatically by the application.
 This is a very common real-life example
Fig.3 shows the algorithms are generated by of why we need the concept like machine
machines learning.

In this picture, the machine is generating the III. USES OF MACHINE LEARNING
algorithm according to the historic data given by
the programmer.  Machine learning in medical industry
 Machine learning in insurance industry
Example2: in the below given figure, the
 Machine learning in action-reaction
machine’s performance is estimated on the basis of
industry.
amount of data analyzes by using machine learning
algorithms and most used algorithms.

Volume 5, Issue 7, July /2018. Page No:539

JASC: Journal of Applied Science and Computations ISSN NO: 1076-5131

With the constant evolution of the machine group data on the basis of their statistical
learning, there is a subsequent rise in the use and characteristics. Further clusters will be given
importance of machine learning. names / labels.

The problems of unsupervised learning are further

divided into a category:

New data  Clustering: This problem is occurred

where we want to discover the groups
or inheritance in the input data E.g.:
Predicated grouping the customers on the
data
behavior of buying things.

prediction

classification
Supervised
Fig.5 showing process of machine learning learning
Machine regression
IV. TYPES OF MACHINE LEARNING learning
A. Supervised Learning: Unsupervised
clustering
learning
This method works by decision giving inputs and
their desired outputs are obtained. In this method,
set of inputs will give corresponding output to
detect errors. Given below is the example of
supervised machine learning showing the input and
giving output. the training information contains
input as well as desired result. This learning is
quite fast and correct in comparison to
Fig.6 Techniques of machine learning
unsupervised. The proper validation is included in
the supervised learning.\
V. CONCLUSION:
The problems of supervised learning are further
Machine Learning is a very vast and big term to
divided into two categories:
study and analyze. This review paper provides only
 Classification: This problem occurs when the outline of machine learning. There are different
the input variable is given a category. E.g.: ways to implement the concept of machine learning
‘red’ or ‘blue’ and ‘diagnosed disease’ and and to get the best results from out there. Basically,
‘non-diagnosed’. machine learning is the concept to analyze the huge
 Regression: the problem of amount of data.
regression occurs when the real value
In past centuries, the maximum amount of time was
is given as a input. E.g.: ‘dollars’ and
let’s say wasted by our programmers in order to
‘weight’.
analyze the huge volume of data on their own and
B. Unsupervised learning: getting the result out of it. We all know that it took
This technique of machine learning to find the hours and hours sometimes even years to estimate
pattern in the given input data by the user. As the data in bulk. When at last the time of results
shown in the below example, the pattern is came, then again they get errors in achieving the
found by the machine and then o/p will be desired output. So this was a very poor and bad
produced. The given model is not aided with approach as it consumes more time and efforts and
all the essential results during the execution also didn’t produced the results according to the
and training. This learning is used to cluster or desire of the user.

Volume 5, Issue 7, July /2018. Page No:540

JASC: Journal of Applied Science and Computations ISSN NO: 1076-5131

This approach also involved team of members and learning is the bright future of science and provides
error detection is one of the major drawbacks of many more technologies to discover and more
this approach. So nowadays, the concept of growth will be there in coming years.
machine learning is introduced. This concept
REFERENCES
familiarizes us with the latest mode of technology
and gives us number of reasons to accept it.
[1] Ankush Sharma, Aakanksha, “A Comparative Study of
Machine learning analyzes the large volume of data Sentiments Analysis Using Rule Based and Support
in very less time and the chances of occurring error Vector Machine”, International Journal of Advanced
are less in this approach. Research in Computer and Communication Engineering
Vol. 3, Issue 3, pp-5898-580, March 2014.
Machine learning enables the machines to develop
[2] Aakanksha, Er.Dinesh Kumar, “A HYBRID
fast and efficient algorithms and data- models for APPROACH FOR TEXT CLASSIFICATION USING
real-time processing of data. It is able to generate HMM, SVM AND GENETIC ALGORITHM”,
accurate and desired results and analysis. International Journal for Technological Research in
Engineering Volume 1, Issue 12, pp-1454-1457, 12,
August-2014
At last I conclude by writing that Machine
Learning is the future of technology and we,
humans will be free by operating machines
anymore. Machines and robots will work even
would think on their own to give results. Machine

Volume 5, Issue 7, July /2018. Page No:541

View publication stats