International Journal of Advanced Research in Engineering and Technology (IJARET)

Volume 11, Issue 6, June 2020, pp. 392-399, Article ID: IJARET_11_06_035
Available online athttp://www.iaeme.com/IJARET/issues.asp?JType=IJARET&VType=11&IType=6
ISSN Print: 0976-6480 and ISSN Online: 0976-6499
DOI: 10.34218/IJARET.11.6.2020.035

© IAEME Publication Scopus Indexed

MACHINE LEARNING IN ARTIFICIAL

INTELLIGENCE
Anirban Chakraborty
M.Tech, Artificial Intelligence,
Lovely Professional University, Punjab, India

Shilpa Sharma
M.Tech, Computer Science and Engineering,
Lovely Professional University, Punjab, India

ABSTRACT
Machine Learning is the core argument and artificial aspect classification of
problem learning and decision making. Scientists therefore introduce machine
learning, often used in the artificial mind. A significant method to automate a wide
variety of activity like human brain is the artificial intelligence preparation
framework. Machine Learning techniques require a planning program to gain search
control of information for different applications automatically. Machine Learning
plays an important role in the field of robotics. It contributes to decision-making and
improves machine efficiency. Machine learning is applied in a large amount of
application. It is the principle concept of intelligence system that helps to introduce
artificial intelligence ingeniously and it also make artificial intelligence very
advanced.
Key words: Machine Learning, Pattern, Bias, Noise, Artificial Intelligence, Computer
Cite this Article: Anirban Chakraborty and Shilpa Sharma, Machine Learning in
Artificial Intelligence, International Journal of Advanced Research in Engineering
and Technology, 11(6), 2020, pp. 392-399.
http://www.iaeme.com/IJARET/issues.asp?JType=IJARET&VType=11&IType=6

1. INTRODUCTION
A category of artificial intelligence that enables software algorithms whose accuracy in
performance prediction without having been specifically programmed is machine learning.
Training is a significant aspect of the analysis of artificial neural networks based on machine
learning.
In order to grasp the simulated world related to the comprehension of computers, various
algorithms are applied in specific machines to prevent the construction of a large computer
with clear programming. A huge amount of collections of data are used to do such analysis to
interpret the outcome. Object detection is the natural method of comparing external

http://www.iaeme.com/IJARET/index.asp 392 [email protected]

Electronic copy available at: https://ssrn.com/abstract=3656899

 Machine Learning in Artificial Intelligence

knowledge of details contained in the brain. The recognition of patterns is closely linked with
top perception. The knowledge and expectation of clustered information are used in both
cases.
Pattern recognition ensures that repetitive features, activities or other attributes are
observed and this is the fundamental way to make sense of the word. Otherwise machine does
constant attempt to identify environmental information which coincides with internal data.
In other words, pattern recognition is a subset of machine learning that is used by various
algorithms for automated decisions.

Figure 1 The Machine Learning Mechanism

2. LITERATURE REVIEW
Literature Survey of some papers which deals with Machine Learning has been done and a
brief summary has been written below:
 In practical learning scenarios, Sally Goldman et Al.[1] presented small amounts of
labeled data with enormous unlabeled data with a joint training strategy to use
unscheduled information to improve the standard supervised learning algorithm. In her
opinion, the division of an instance space is defined by two types of hypothesis (eg.
space for instance in the decision tree partition with one equivalent tree class.) She
concluded that two supervised learning algorithms can successfully label each others
data.
 Zoubin Ghahramani et.al.[2] presented a description of unregulated learning through
mathematical models. He argued that uncontrollable thinking may be driven by
analytical knowledge and Bayesian concepts. Furthermore, he argues that statistics
provide a consistent framework for data and logic analysis and he discloses
frameworks such as graphic designs that play a vital role in learning the use of various
data types.
 Rich Caruana et al[3] explored the similarity of ten supervised forms of learning over
the last decade with supervised process. SVMS include neural networks, regression of
logistics, naive bays, memory-driven computing, random wood, decision-making
plants, vine and stumbling. They examined the effect of calibrating models via plate
scaling and isotonical regression to evaluate the learning methods.
 Niklas lavesson et.al[4] stated that efficiency is always calculated only in terms of
accuracy. However, several researchers have taken a specific method to assess

http://www.iaeme.com/IJARET/index.asp 393 [email protected]

Electronic copy available at: https://ssrn.com/abstract=3656899

 Anirban Chakraborty and Shilpa Sharma

controlled learning via cross-validation experiments. i.e. the measurement function is

that they can handle only the spaces of two dimensional instances. They presented the
design and implementation and demonstrate use with a series of experiments of a
widespread multidimensional measuring function. The results show that there can be
cases for which measurement functions are in the possibility of capturing aspects of
performance that can not be captured with cross validation tests.
 Yogowati Praharsi et.al[5] proposed that three supervised methods of learning are
approached by him with the neighbour (k-NN) algorithm. The description of
supported vector data and the support vector machine(SVM) introduced a new class
which is not difficult to study and is not used in description and classification of data.
The output shows that the selection feature can be considered a replacement to the
forward selection based on mean information gain and a standard deviation threshold.
 In paper [6], the authors take machine learning into account and discuss its trends and
future. In the world of today, the technical fields are growing fast at the crossroads
informatics and statistics and at the heart of artificial technology of knowledge and
intellect. In the development of machine learning, many new algorithm of searching is
used as it ensures principle and current explosion of online data availability at low-
cost calculation.
 In paper [7], the authors introduce the machine learning tools for research purpose
where the background is concerned as Machine scope and application information as
Ecological analytical learning methods. Various machine learning techniques to
evaluate complexity of a dataset are implemented.

3. PROBLEM FACED IN LEARNING

Because too many choices are taken, learning relies on machine-to-machine and algorithms
are regarded as a dynamic operation. From an interpretation of a question to a solution, too
many problems arises and it is difficult to respond to a system. The system will always
address different forms of problems and conditions. Although different inputs produce
different outputs, only the optimized and appropriate output should be taken into account.

3.1. Problems Faced during Learning Process are as follows:

Bias
Some errors in learning algorithms are considered prejudices. At the same time the question
occurs when two error sources are reduced that prohibit supervised learning algorithms.
Noise
In real life circumstances, unintended data and data imperfection are now normal. The noise
in the data is degrading the learning process but one of the characteristics of the learning
algorithm is the use of all valid data.
Pattern Recognition
The concept of pattern recognition is that all inputs are replied correctly, and the matching
procedure for all inputs is done and carried out according to statistical adjustments.
The machine, because of its mathematical models, is well-known (square, rectangle,
circle, etc), but the processing of these inputs is also different for the machine.
Inputs and outputs are both perceived through supervised learning. The algorithm must
generate all inputs in response to all the inputs. all training data from supervised learning must
be generated by the algorithm.

http://www.iaeme.com/IJARET/index.asp 394 [email protected]

Electronic copy available at: https://ssrn.com/abstract=3656899

 Machine Learning in Artificial Intelligence

When an individual provides instant input, guided intervention learning takes place. Some
steps have to be taken to resolve any given problem through supervised learning:
1. Definition and form assessment in training
2. Training set for collection
3. Knowledge of the role of the feedback
4. Determining the learning structure
5. The decision to perform a learning algorithm based on collected data is completed
6.Again the training set should be measured and the accuracy of the learned function and
performance is to be measured.

Table 1 Pattern Recognition

Figure 2 Supervised Learning Algorithm

Two types in learning exist
1. Grouping of responses with only meanings of facts (true or false).
2. Recovery of real values of responses.
Controlled inputs for learning are obtained in Supervised Learning, however the desired
outcomes and environmental mitigation have not been controlled. Since it has struggled, a
structured dataset should be developed for unsupervised learning such as clustering and
dimensional reduction.

http://www.iaeme.com/IJARET/index.asp 395 [email protected]

Electronic copy available at: https://ssrn.com/abstract=3656899

 Anirban Chakraborty and Shilpa Sharma

Figure 3 Working Mechanism of Supervised Algorithm

4. HIERARCHIAL CLUSTERING
Hierarchy is a cluster analysis method, in which we look forward to build a cluster hierarchy.
The purpose of this algorithm is to construct a multi-level cluster hierarchy tree by
constructing a cluster tree.
Inputs: objects represented as vectors
Output: a group of associations represented as a “ Dendogram”
Algorithm
 hclust (F,:set of instances ):tree
 var:K:/*set of clusters*/
 N/*matrix containing distance between 2 clusters*/
 For each fԐF}do
 Make b as leaf node in K
 Done
 For each pair b,dԐK do
 Ma,d←f(b,d)
 Done
 While( not all instances in one cluster)do
 Find the most similar pair of cluster in N
 Merge these two cluster into one cluster
 Update N to reflect the merge operation
 Done
 Return K

5. K-MEANS CLUSTERING
A vector quantization method for cluster analysis called k-means clusters, is used in data
mining. The goal of k-means is to divide n observations as a test into the nearest cluster.

http://www.iaeme.com/IJARET/index.asp 396 [email protected]

Electronic copy available at: https://ssrn.com/abstract=3656899

 Machine Learning in Artificial Intelligence

Algorithm
 K-means ((X={f1…..fn} Rm,k):2J)
 C:2J /* μ a set of clusters*/
 f:JxJm->J/*distance function*/
 μ:2J->J/*μ computers the mean of a cluster*/
 select C with k initial centers v1,….vk
 while stopping criterion not true do
 for all clusters cH Ԑ C do
 cH←{fi\ᵾv1f(di,vH) ≤f(fi,v1)}
 done
 for all means vH do
 Hv←μ(cH)
 done

Figure 4 Working Mechanism of Unsupervised Learning Model

6. CONCLUSION
Studies have been carried out to evaluate the performance of learning algorithms. It is the
interesting problem in many ways. Few problems, such as the study of assessment processes
and the criteria that tests output and a function that structurally defines the procedures. The
outcome of the research is that the classification outcomes may be correctly measured as in
the cross validity check. The classification output analysis is important since the approach
does not operate as a process that returns the answer. Measurement-based evaluation for
classification performance was also examined and results from factual experiments are
provided, which enhance previous publication of theoretical arguments for measurement-
based assessment. This experiment was able to distinguish between classifiers that were
accurately and complexly acquired.

ACKNOWLEDGEMENT AND FUNDING

Proper funding has been received.
We, the authors are very much thankful to Mr. Arun Kumar Chakraborty for funding our
research.

http://www.iaeme.com/IJARET/index.asp 397 [email protected]

Electronic copy available at: https://ssrn.com/abstract=3656899

 Anirban Chakraborty and Shilpa Sharma

REFERENCES
[1] Philip Adler, Casey Falk, Sorelle A. Friedler, Tionney Nix, Gabriel Rybeck, Carlos
Scheidegger, Brandon Smith, and Suresh Venkatasubramanian. (2018). Auditing Black-
box Models for Indirect Influence. Knowledge and Information Systems 54, 1 (Jan. 2018),
95–122.

[2] Julia Angwin and Jeff Larson. (2016). Machine Bias. ProPublica. (May 2016).
https://www.propublica.org/article/machine-bias-risk-assessments-incriminal-sentencing.

[3] Edmond Awad, Sohan Dsouza, Richard Kim, Jonathan Schulz, Joseph Henrich, Azim
Shariff, Jean-François Bonnefon, and Iyad Rahwan. (2018). The Moral Machine
Experiment. Nature 563, 7729 (Nov. 2018), 59–64.

[4] Reuben Binns, Max Van Kleek, Michael Veale, Ulrik Lyngs, Jun Zhao, and Nigel
Shadbolt. (2018). “It’s Reducing a Human Being to a Percentage”; Perceptions of Justice
in Algorithmic Decisions. In Proc. CHI.

[5] Alexandra Chouldechova. (2017). Fair Prediction with Disparate Impact: A Study of Bias
in Recidivism Prediction Instruments. Big data 5, 2 (2017), 153–163. arXiv: 1610.07524.

[6] Anupam Datta, Shayak Sen, and Yair Zick. (2016). Algorithmic Transparency via
Quantitative Input Influence: Theory and Experiments with Learning Systems. In Proc.
IEEE S&P.

[7] William Dieterich, Christina Mendoza, and Tim Brennan. (2016). COMPAS Risk Scales:
Demonstrating Accuracy Equity and Predictive Parity. Northpointe Incorporated. (2016).
http://go.volarisgroup.com/rs/430-MBX-989/images/
ProPublica_Commentary_Final_070616.pdf.

[8] Michael Feldman, Sorelle Friedler, John Moeller, Carlos Scheidegger, and Suresh
Venkatasubramanian. (2014). Certifying and Removing Disparate Impact. arXiv. (2014).
http://arxiv.org/abs/1412.3756.

[9] Nina Grgić-Hlača, Elissa M. Redmiles, Krishna P. Gummadi, and Adrian Weller. (2018).
Human Perceptions of Fairness in Algorithmic Decision Making: A Case Study of
Criminal Risk Prediction. In Proc. WWW.

[10] Nina Grgić-Hlača, Muhammad Bilal Zafar, Krishna P Gummadi, and Adrian Weller.
(2016). The Case for Process Fairness in Learning: Feature Selection for Fair Decision
Making. In Proc. NIPS Symposium on Machine Learning and the Law.

[11] Nina Grgić-Hlača, Muhammad Bilal Zafar, Krishna P Gummadi, and Adrian Weller.
(2018). Beyond Distributive Fairness in Algorithmic Decision Making: Feature Selection
for Procedurally Fair Learning. In Proc. AAAI.

[12] Moritz Hardt, Eric Price, and Nathan Srebro. (2016). Equality of Opportunity in
Supervised Learning. In Proc. NIPS.

[13] Kenneth Holstein, Jennifer Wortman Vaughan, Hal Daumé III, Miro Dudik, and Hanna
Wallach. (2019). Improving Fairness in Machine Learning Systems: What Do Industry
Practitioners Need?. In Proc. CHI.

[14] Farnaz Jahanbakhsh, Wai-Tat Fu, Karrie Karahalios, Darko Marinov, and Brian Bailey.
(2017). You Want Me to Work with Who?: Stakeholder Perceptions.

http://www.iaeme.com/IJARET/index.asp 398 [email protected]

Electronic copy available at: https://ssrn.com/abstract=3656899

 Machine Learning in Artificial Intelligence

[15] Kodimalar Palanivel and Chellammal Surianarayanan, (2019) An Approach for Prediction
of Crop Yield Using Machine Learning and Big Data Techniques, International Journal of
Computer Engineering and Technology, 10 (3), pp.110-118.

[16] B. M. G. Prasad and P. V. S. Sreenivas, (2019) An Implementation of Software Effort

Duration and Cost Estimation with Statistical and Machine Learning Approaches,
International Journal of Computer Engineering and Technology,10(1), pp. 81-93

[17] Kushal Jani, Punit Lalwani, Deepak Upadhyay, Dr. M.B. Potdar, (2018) Performance
Evolution of Machine Learning Algorithms for Network Intrusion Detection System.
International Journal of Computer Engineering and Technology, 9(5), pp. 181-189.

[18] Temesgen Abera Asfaw, (2019) Prediction of Diabetes Mellitus Using Machine Learning
Techniques, International Journal of Computer Engineering and Technology10(4), pp.25-
32

AUTHORS DETAILS

Mr. Anirban Chakraborty

M.Tech, Artificial Intelligence,
Lovely Professional University,
Phagwara, Punjab, India

http://www.iaeme.com/IJARET/index.asp 399 [email protected]

Electronic copy available at: https://ssrn.com/abstract=3656899