University Report: Machine Learning for Fraud Detection, MITS6011

This report, focusing on machine learning for fraud detection, examines the potential threats organizations face and how machine learning can play a crucial role in mitigating them. It begins with a literature review discussing how organizations miss potential threats, often due to vulnerabilities in their IT systems, including spam messages, and inefficient security measures like firewalls and antivirus software. The report then explores how machine learning, particularly anomaly detection, can be applied to identify fraudulent activities and improve security. It highlights the benefits of machine learning, such as its ability to analyze data, detect patterns, and identify suspicious activities. The research methodology section outlines the approach used to analyze data and assess the role of machine learning in fraud detection, including considerations of ontology, epistemology, and positivism. The report provides valuable insights into how organizations can leverage machine learning to enhance their cyber security operations and protect against various cyber attacks.

Running head: MACHINE LEARNING FOR FRAUD DETECTION
MACHINE LEARNING FOR FRAUD DETECTION
Name of the Student
Name of the University
Author note

MACHINE LEARNING FOR FRAUD DETECTION
Table of Contents
Literature review: 1
How potential threats are missed by the organization: 1
Machine learning can play a crucial role in detection of frauds: 3
Research Methodology: 6
Reference: 10

MACHINE LEARNING FOR FRAUD DETECTION
Literature review:
How potential threats are missed by the organization:
The major threat to the organization that is responsible for bringing down the working
structure of the business organization due to serious attacks that has caused downfall are
labelled as potential threats for an organization. The most potential threat to an organization
happens on the most vulnerable and sophisticated area of the organization. There are severe
threats to the organization that are arising due to the spam messages that are sent to the
employee’s accounts in the email profiles. Accessing those links and providing the private
information in the links cost the organization in a vulnerable way. The attackers get access to
the private details of the organization and also are able to access the network server of the
organization and hence these are the main reason behind the different types of the malicious
attacks in the organizations and this has been clearly stated by the author Agrawal and
Agrawal (2015). The data of the attacks or the area from where the attacks are originating are
often not available to the organization and hence the organization is unable to track the areas
that are causing the harm. The detection formats that are used by the organizations are poor.
The small loopholes that are created due to the inefficient security of the IT systems of the
organizations are the main reasons that are causing the potential threat to the systems of the
organizations. The firewall systems or the anti viruses used by the organizations need to be
more efficient [15]. The up to date cyber security systems or the firewalls are required for
avoiding the problems that are causing the security malfunction in the organization systems.
Malware is software that is malicious. It is the principal cyber-assault weapon and
contains viruses, worms, Trojans, ransomware, administrative devices, spyware bots, bugs
and kits. When a user clicks a connection or acts, it is installed. Malware can prevent access
to data and programs within, obtain information and infiltrate systems.

MACHINE LEARNING FOR FRAUD DETECTION
As per the statement of the author, Mena (2016), The significance of addressing
Internet safety has been understood by most companies. With almost 2/3 of a group of little
organisations lately surveyed that have suffered cyber attacks in the past two years, there is
increasing talk about the hazards of a lack of cyber security. The following dangers include:
Private data compromise: These days companies depend strongly on the information
they collect, whether it be market information, multiple account details or customer private
data. If a cyber hack happens, this information not only can be robbed by another
organization, but also the data can be changed so that operational reliability can be drastically
damaged by the company.
Costly costs for recovery. Not only does a violation of safety jeopardize data, it can
have catastrophic consequences. Most of these are "concealed" expenses, which could affect
your company for up to two years following the incident [16]. The loss of both moment and
money, regardless of whether it is fresh IT-training, the acquisition of fresh software, or the
extensive process of restoration of lost information.
The organization systems have become vulnerable to the cyber attacks. The security
systems of the organizations are not efficient enough to track the vulnerabilities that are done
by the personal system of the employees of the organization. The monitoring systems that are
needed to be implemented around the network system of the organization are lot present in
the organization. This result in being prey to the attacks that are done by the hackers or all
kind of fraud related attackers and brings done the organization business infrastructure
causing immense loss to the system.
Machine learning can play a crucial role in detection of frauds:
According to the author Ayoubi et al. (2018), the machine learning is referred to the
application of the IT advancement which is known as the Artificial Intelligence. The

MACHINE LEARNING FOR FRAUD DETECTION
application helps in providing the ability to the systems where the systems are able to directly
communicate with the help of learning. This feature directly helps in improving the level of
experience without the intervention of the explicit based programs. The main focus of the
machine learning attribute is on the development of the computer based programs that are
able to have full access of data and also utilize them for learning by the system.
The machine learning has ample of attributers that makes it different from any other
innovation in the digital world. The machine learning an be referred to as the process which
analyses the data for all kind of analytical work on data execution. They work on the AI
infrastructure and help to obtain the data along with the patterns that helps the organization
commonly to make decisions on the critical areas like the detection of the frauds. The
machine learning attribute which is known as the Anomaly Detection is the process which
enables the classification of the rare terms, events or the detection based observation on the
suspicious data [12].
It can be derived from the current scenario of all the organizations that the complexity
on the networks have increased due to the high scalability of the networks and has proved to
be inefficient for the manual administration process. The main harm that has been caused to
the network security of the organizations is the human error or the error of the employees of
the organizations well as the management system of the organization [14]. The vulnerabilities
caused by the employees of the organization has pushed the organization to shift the design
pattern to the use of machine learning for the purpose of the detection of the fraud activities
that are affecting the business infrastructure of the organization. They are also responsible for
detection of the areas that are initiating the fraudulent activities or the areas that are
vulnerable to the attackers.

MACHINE LEARNING FOR FRAUD DETECTION
After the statement of Yaram (2016), it has been deduced that the organizations that
are based on the marketing of the goods and products directly to the customer and the
business organizations that deal with transactions in a daily basis are prone to the attacks in
the current scenario. The world has been adopting the digitalization in a rapidly increasing
nature. The digital payment systems includes the sharing of the credentials on the internet
based login pages. Even the credit card and the debit card companies are also facing a huge
growth in the sales as people are utilizing the facility on a huge scale [2]. The large scale of
usability also implies about the large scale of vulnerabilities that may cause and are
happening with the cyber security issue on these company where the valuable information are
often set on stake due to less secured system and the attacks on those systems of the
organizations. The Machine learning provides efficient fraud detection system usable to
detect and sense the fraud. The high accuracy and high efficiency of the ML focuses on the
fraudulent transactions of the organizations and also detects the PHISHING activities from
the node it is being carried out.
The ML has a huge impact on the detection process of the fraudulent activities and
provides restrictions to the usability, but in this course, the genuine users are sometime
prevented from having the access to the organization network or the access to the payment
related sources. Therefore it can also be deduced that a large number of negative impacts of
the usage of the ML can turn the area into a negative customer feedback for the organization
[3].
The machine learning can play a crucial role in detection of the potential threats in the
organization as they provide Fraud management data, assists in Configuration management,
performance based management, and security management for the organization.

MACHINE LEARNING FOR FRAUD DETECTION
The machine learning consists of network monitoring for well-established threat
patterns. This makes the network susceptible to zero-day assaults, however. This is critical
when fresh attacks arise on a daily basis. There is a clear need for solid safety measures and
ML's role in this area has been thoroughly investigated. There have also been attempts to use
ML to detect misuse so that complicated attack patterns from historical information can be
obtained and generic rules to recognize variants of known attacks can be generated. Zero day
attacks have also been investigated for anomaly detection using ML. This involves learning
models and detecting deviations from the normal behaviour. The above attempts demonstrate
promising outcomes for cognition to be incorporated into network management [5].
Leveraging ML alone for the various network leadership functions won't satisfy the cognitive
management vision. In reality, a cognitive control circuit is necessary which the part of the
Machine learning is.
Research Methodology:
The utilization of the efficient and appropriate type of procedures along with the
techniques for the research based methodology has helped the researchers to obtain all the
value based information that are detailed i8jn nature. The research had been done in a very
detailed format for deducing how potential threats are missed by the organization And how
machine learning can play a crucial role in detection of these frauds. We have developed an
approach to improve the cyber security operations within the organization by using the
Machine Learning approach. The article or the chapter has successfully identified and
classified the significant points that are referred to as the methods and the process for
evaluating and revising the data for the article [7]. The main concept and the process for the
analysis of the data on the implementation of the Machine Learning in the organization for
the purpose of detection of the errors and the fraudulent activities that are happening in the

MACHINE LEARNING FOR FRAUD DETECTION
organization. As mentioned in the title, the study methodology for the thesis is covered in this
section. The writer describes the research plan, the research method, the research approach,
the information collection techniques, sample choice, the research process, and the type of
information assessment, the ethical factors and study constraint for the project has been
executed in more and accurate details.
The utilization of the accurate and specified detailed procedure for performing the
research has been procured of the research based philosophy that discloses the procedure that
helps in the study of the research on the role of machine learning on the fraud detection of the
organizations.
The philosophy of research is an significant component of the methodology of
studies. Ontology, epistemology and axiology are categorized in research philosophy. These
philosophical methods allow the researchers to decide which strategy they should adopt and
why. There are significant assumptions in the philosophy of studies which explain the views
of the scientist on the globe. The study approach and the techniques of this approach will be
based on those assumptions
ONTOLOGY: The nature of truth is a basis for Ontology. It is categorized on
objective and subjective grounds. The first element of ontology, objectivism, shows that in
fact social objects persist outside of the social actors. The social phenomena emerging of the
perceptions and implications of these social actors concerned with their presence are worried
with subjectivity.
EPISTEMOLOGY: The understanding of an appropriate field of research is
considered epistemology. It can be split into two dimensions: resource scientist, and
sensationalist. The resource scientist deals with the information from the natural scientist's

MACHINE LEARNING FOR FRAUD DETECTION
view. On the other side, the' feeling investigator ' concerns employees with regard to their
managers ' emotions and attitudes.
POSITIVISM: The natural scientist's philosophical method is noted as positivism as
the natural scientist's job relies on observed social entities. On the grounds of information
collection and hypothesis creation, a research plan is approached. These hypotheses are tested
and validated for further investigation. The positive researchers follow extremely organized
methods to promote the hypothesis as another element of this philosophy. In addition,
positivism operates on quantifiable observations and consequently statistical analysis is
achieved.
The most suitable research philosophy for identifying processes and methods used to
detect the fraud activities and the source and the reason of the fraud activities in the
organization by the cyber-attacks has been chosen for positivism in this study. In addition, the
use of positive activity has ensured that resources leading to the use of the Machine Learning
which has been evaluated as the method implication for the detection and rectification of the
cyber-attacks in the organizations that have brought down the business performance of the
organization [8]. The implementation of information collection and analysis has been
demonstrated by Positive philosophy or positivism.
Research Approach provides the format needed to collect and analyze the information
to be followed. Deductive strategy was detected as best suited to detect the fraud and
loopholes in the organization. The use of Deductive promotes hypothesis creation and
determines the requirements for selecting and rejecting hypotheses [9]. In addition, the
hypothesis considered during the studies was justified by the deductive strategy.
The research design demonstrated that the development of the comprehensive data
recovery and interpretation process framework was permitted. In this study, the analyst chose

MACHINE LEARNING FOR FRAUD DETECTION
Conclusionary studies to collect appropriate information the organization to detect fraud by
the implementation of the machine learning [11]. The implementation of conclusive design
can explain the method for the collection and interpretation of raw information to conclude
the research conducted.
The implementation of conclusive design can explain the method for the collection
and interpretation of raw information to conclude the research conducted. The
implementation of conclusive design enables the method for collecting raw information and
interpreting the completion of the undertaken research to be explained.
Chappell (2015) asserted that data offers the necessary information during the conduct
of the studies to help conclude and advise the organization on the matter of detection of fraud
and the use of machine leaerning. The main and secondary information sources were assessed
in this research. The theoretical context for this research was the implementation of
secondary data. The thorough literature review was used as a secondary source of
information. On the other side, the main use of information enabled the investigator to obtain
raw information and to develop the research context. The investigator chose 20 staff and 80
clients for the information collection procedure in connection with the organization. The
employee’s answers were collected via an internet survey form. In order to select the staff for
this online survey,’ random likelihood samples methods' have also been used.

MACHINE LEARNING FOR FRAUD DETECTION
Reference:
[1] S. Ayoubi, N. Limam, M.A. Salahuddin, N. Shahriar, R. Boutaba, F. Estrada-Solano and
O.M. Caicedo, Machine learning for cognitive network management. IEEE Communications
Magazine, 56(1), pp.158-165, 2018.
[2] A.C. Bahnsen, D. Aouada, A. Stojanovic and B. Ottersten, Feature engineering strategies
for credit card fraud detection. Expert Systems with Applications, 51, pp.134-142, 2016.
[3] J.O. Awoyemi, A.O. Adetunmbi and S.A. Oluwadare, Credit card fraud detection using
machine learning techniques: A comparative analysis. In 2017 International Conference on
Computing Networking and Informatics (ICCNI) (pp. 1-9). IEEE, 2017, October.
[4] S. Agrawal and J. Agrawal, Survey on anomaly detection using data mining
techniques. Procedia Computer Science, 60, pp.708-713, 2015.
[5] A.O. Adewumi and A.A. Akinyelu, A survey of machine-learning and nature-inspired
based credit card fraud detection techniques. International Journal of System Assurance
Engineering and Management, 8(2), pp.937-953, 2017.
[6] J. Mena, Machine learning forensics for law enforcement, security, and intelligence.
Auerbach Publications, 2016.
[7] A. Dal Pozzolo, G. Boracchi, O. Caelen, C. Alippi and G. Bontempi, Credit card fraud
detection and concept-drift adaptation with delayed supervised information. In 2015
international joint conference on Neural networks (IJCNN) (pp. 1-8). IEEE, 2015, July.
[8] A. Roy, J. Sun, R. Mahoney, L. Alonzi, S. Adams and P. Beling, Deep learning detecting
fraud in credit card transactions. In 2018 Systems and Information Engineering Design
Symposium (SIEDS) (pp. 129-134). IEEE, 2018, April.

MACHINE LEARNING FOR FRAUD DETECTION
[9] A. Dal Pozzolo, Adaptive machine learning for credit card fraud detection, 2015.
[10] S. Yaram, Machine learning algorithms for document clustering and fraud detection.
In 2016 International Conference on Data Science and Engineering (ICDSE) (pp. 1-6).
IEEE, 2016, August.
[11] S. Subudhi, and S. Panigrahi, Quarter-sphere support vector machine for fraud detection
in mobile telecommunication networks. Procedia Computer Science, 48, pp.353-359, 2015.
[12] S. Safavi, H. Gan, I. Mporas, and R. Sotudeh, Fraud detection in voice-based identity
authentication applications and services. In 2016 IEEE 16th international conference on data
mining workshops (ICDMW) (pp. 1074-1081). IEEE, 2016, December.
[13] D. Chappell, Introducing azure machine learning. A guide for technical professionals,
sponsored by microsoft corporation, 2015.
[14] G. Rushin, C. Stancil, M. Sun, S. Adams and P. Beling, Horse race analysis in credit
card fraud—deep learning, logistic regression, and Gradient Boosted Tree. In 2017 Systems
and Information Engineering Design Symposium (SIEDS) (pp. 117-121). IEEE, 2017, April.
[15] B. Coma-Puig, J. Carmona, R. Gavalda, S. Alcoverro, and V. Martin, Fraud detection in
energy consumption: A supervised approach. In 2016 IEEE International Conference on
Data Science and Advanced Analytics (DSAA) (pp. 120-129). IEEE, 2016, October.
[16] O.S. Yee, S. Sagadevan, and N.H.A.H. Malim, Credit card fraud detection using
machine learning as data mining technique. Journal of Telecommunication, Electronic and
Computer Engineering (JTEC), 10(1-4), pp.23-27, 2018.