Research Report 2019

Artificial intelligence in fraud detection and prevention

In recent years, due to the expansion of financial institutions, as well as the popularity of the World
Wide Web and e-commerce, a significant increase in the volume of financial transactions observed.
Along with the increase in trading volume, it had seen a huge increase in the number of cheats by
abusive users, resulting in billions of dollars annually causing losses worldwide. So it was necessary
to consider the identification and diagnosis of the fraud techniques.

Fraud is a dominant form of white collar crime that continues to extract a significant toll not only on
the organizations, but also on investors, financial institutions, and the economy in general. Reserve
Bank of India has defined fraud as “All instances wherein Banks have been put to loss through
misrepresentation of books of accounts, fraudulent encashment of instruments like cheques, drafts and
bills of exchange, unauthorized handling of securities charged to banks, misfeasance, embezzlement,
theft, misappropriation of funds, conversion of property, cheating, shortages, irregularities etc.”
Examples of fraud include insurance fraud, credit card fraud, accounting fraud, etc.

A financial institution is a company engaged in the business of dealing with financial and monetary
transactions such as deposits, loan and investment and currency exchange. Financial institution is also
known as banking institution. According to a Delloite survey, 93% respondents indicated that there
has been an increase in fraud incidents in the banking institution in the last two years.

There are dozens of ways in which an individual can commit banking fraud. Typically, fraud in
financial institution can be categorized into 3 main categories: Corruption, Asset Misappropriation and
Financial Statement Fraud. Corruption includes cases of conflict of interest, bribery, illegal gratuities
and extortions. Asset misappropriation may be embezzlement or inventory related fraud. Financial
statement fraud involves overstating or understating the assets or revenue generated.

Fraud cases involving theft of identity are a serious and growing problem in the era of internet banking.
With so many transactions done online, hackers have the ability to frequently access bank account and
credit card information from unwitting consumers. Fraudsters can also use obtained names and
addresses to apply for fraudulent accounts, credit cards and loans.

Globalization, technology advancement and other factors, have led to increased cases of finacical
fraud. In addition, frauds have no constant patterns and they always change their behavior over time.

Artificial intelligence has been growing at a rapid rate over the past couple of years,

1
susmita dhital(HND/3rd Semester)
 Research Report 2019

The field has been growing at a rapid rate over the past couple of years,

Artifical intelligence has reach the mountain in the todays world.

Traditional methods of data analysis have long been used to detect fraud. They require complex and
time-consuming investigations that deal with different domains of knowledge like financial,
economics, business practices and law.

Primarily, most banks employ Rule-based Systems with manual evaluation for detecting fraud.
Although these systems were doing a pretty decent job, in the recent years, they have become more
inconsistent. That’s because new fraud patterns are evolving rapidly and these systems are unable to
evolve accordingly, allowing frauds to go undetected, and resulting in huge financial losses.

Considering all these challenges and shortcomings, Machine Learning can play a vital role in effective
and efficient fraud detection in the banking industry.

Data Mining and Computational Intelligence techniques are commonly used in fraud detection. Here, I
will be mainly focusing on Credit Card Fraud Detection and talk about the techniques, approaches and
lessons learnt, based on our experience in building a Machine Learning application for detecting credit
card fraudulent transactions in real time.

https://medium.com/engineered-publicis-sapient/fraud-detection-in-banking-industry-and-
significance-of-machine-learning-dfd31891a0b4

Summary and Observations

 We compared the performance of all the 4 algorithms we implemented and found that neural
network technique gave the best results. Finally, we created a Rest API in Flask for real-time fraud
prediction using the feedforward neural network model we trained. The assessment/benchmarking
and comparisons were done against F1 metrics score (using Precision/Recall).

 While detecting real-time fraudulent credit card transactions, it is tricky to flag a transaction as
fraud, since usually, fraudulent transactions don’t happen in one big transaction, but across multiple
smaller transactions over a period of time. In such cases, a fraud score/weightage-based approach
should be used. This will help detect transaction-level or account-level frauds more effectively and
accurately.

2
susmita dhital(HND/3rd Semester)
 Research Report 2019

Conclusion

From our learning, we found ML-based systems are better equipped to detect fraud, as they are able to
detect and recognize thousands of patterns as opposed to rule-based systems. In the banking industry,
we are definitely witnessing more and more banks embrace ML for fraud detection. A model with just
the supervised ML algorithms/techniques would not be effective enough to detect fraud efficiently and
provide the right, required insights. I believe, a model with the right combination of supervised and
unsupervised ML techniques would enable banks better to detect frauds more accurately.

I still believe we have some more way to go when it comes to leveraging unsupervised learning
techniques for fraud detection. More detailed research needs to be done with different ML techniques
by considering aspects like their performance, hyper-parameter optimization and operationalizations.

3
susmita dhital(HND/3rd Semester)
 Research Report 2019

Detection of fraudulent activity is thus critical to control these

costs. This paper hereby addresses bank fraud detection via

the use of data-mining techniques; association, clustering,

forecasting, and classification to analyze the customer data in

order to identify the patterns that can lead to frauds. Upon

identification of the patterns, adding a higher level of

verification/authentication to banking processes can be added

Keywords: Data mining techniques, banking sector, fraud, and

authentication.

4
susmita dhital(HND/3rd Semester)
 Research Report 2019

A good fraud detection system should be able to detect the transaction in real time and with

accuracy. There are two kinds of fraud detection systems: anomaly detection and misuse detection.

Anomaly detection does not need to be trained whereas misuse detection systems needs a training

phase. We are going to focus in this paper in anomaly detection systems.

Several approaches in terms of machine learning (ML) have been implemented. Typical ML

algorithms used are KNN (K Nearest Neighbor), decision trees and logistic regression, but they are

supervised methods, meaning that they need to learn by labels in order to understand what

transactions are fraudulent or not. If the company lacks this information, these algorithms cannot

be trained. Figure 1 shows a typical process for fraud detection where ML or Deep Learning (DL)

algorithms can be used.

There are many issues that make effective fraud management a challenging task. These include:
enormous and ever-expanding volumes of data, the growing complexity of systems, changes in
business processes and activities and continuous evolution of newer fraud schemes to bypass existing
detection techniques. Detecting fraudulent financial statements is a difficult task when using normal
audit procedures due to limitation in understanding the characteristics of financial statements, lack of
experience and dynamically changing strategies of fraudsters.

It is a part of operational risk and has been classified as Internal and External fraud. Internal Fraud is
the risk of unexpected financial, material or reputational loss as the result of fraudulent action of
persons internal to the firm. Losses are due to acts of a type intended to defraud, misappropriate
property or circumvent regulations, the law or company policy, excluding diversity/discrimination
events, which involves at least one internal party. It includes misappropriation of assets, tax evasion,
intentional mismarking of positions, bribery.

Reserve Bank of India has defined fraud as “All instances wherein Banks have been put to loss through
misrepresentation of books of accounts, fraudulent encashment of instruments like cheques, drafts and
bills of exchange, unauthorized handling of securities charged to

5
susmita dhital(HND/3rd Semester)
 Research Report 2019

Therefore, fraud detection techniques to identify users motivated to do a lot of scientific research. To
check the volume of such information data mining techniques used. Data mining, knowledge discovery
process that finding patterns in large data sets by combining techniques from statistics, artificial
intelligence, database management and etc. In data mining techniques we examine and analyze
information to discover uncertain communication and hidden patterns of data. This study aimed to
detect fraud in banking data using classification algorithms.

and this a motivation for doing many researches(1). Data mining techniques and classification
algorithms are used to examine the large amount of information and discovery of data. These methods
are typically classified into two groups: Anomaly Detection and Abuse Detection Method. In the first
method, customer behavior history

Fraud detection is the recognition of symptoms of fraud where no prior suspicion or tendency to fraud
exists. Examples include insurance fraud, credit card fraud and accounting fraud.

6
susmita dhital(HND/3rd Semester)
 Research Report 2019

We are living in a world which is rapidly adopting digital payments systems. Credit card and payments
companies are experiencing a very rapid growth in their transaction volume. In third quarter of 2018,
PayPal Inc (a San Jose based payments company) processed 143 billion USD in total payment volume
[4]. Along with this transformation, there is also a rapid increase in financial fraud that happens in
these payment systems. An effective fraud detection system should be able to detect fraudulent
transactions with high accuracy and efficiency. While it is necessary to prevent bad actors from
executing fraudulent transactions, it is also very critical to ensure genuine users are not prevented from
accessing the payments system. A large number of false positives may translate into bad customer
experience and may lead customers to take their business elsewhere. A major challenge in applying
ML to fraud detection is presence of highly imbalanced data sets. In many available datasets, majority
of transactions are genuine with an extremely small percentage of fraudulent ones. Designing an
accurate and efficient fraud detection system that is low on false positives but detects fraudulent
activity effectively is a significant challenge for researchers. In our paper, we apply multiple binary
classification approaches - Logistic regression , Linear SVM and SVM with RBF kernel on a labeled
dataset that consists of payment transactions.Our goal is to build binary classifiers which are able to
separate fraud transactions from non-fraud transactions. We compare the effectiveness of these
approaches in detecting fraud transactions.

7
susmita dhital(HND/3rd Semester)
 Research Report 2019

Artificial Intelligence (AI) is a significant advancement in the technology that has everyone talking on
its exciting promises in the technology world. When it comes to AI, it also includes its areas such as
machine learning (ML) and profound learning. While AI could be described as the ability of machines
to make intelligent human-like decisions and improve over time, ML involves building models, mostly
statistical models that give predictive results and can be developed.

Many who are not very informed about this area associate AI with robots that imitate human
operations and intelligence in their working. This is of course since film industries have taken their
time in convincing us AI refers to robots. While this is true, there is more to it from face recognition,
fingerprints recognition, chatbots, pattern recognition, predictive business models and sentimental
analysis. Previously, AI integration in the software development was only possible to the large
companies that had the resources to hire highly qualified professionals. Over time, AI frameworks with
high abstraction level have been developed, and with few coding lines in any programming language
of choice, one can be able to come up with an intelligent system.

The main aim of this thesis is to investigate how AI is being implemented in different industries,
particularly in the field of finance. This thesis will discuss the following research questions: What are
the opportunities and challenges that come with AI? What are the uses of AI in financial sector? How
is AI adopted globally and in Finland? What impact will AI have in the jobs and functions of finance
profession? How does the future of AI look like? What steps should organizations take to succeed in
the competitive era of technology? Document analysis of qualitative research method will be used as
an approach to be solving these questions.

A lot of modern smartphone devices can recognize faces and fingerprints and use that as a primary
security check. AI integration has also been used in the banking system in fraud detection where they
spot abnormal operations and raise the alarm. In the business world, many business applications today
have been built using AI in producing predictive models that could be improved over time. The
predictive models could be integrated into real estates prices estimations, stock exchange prices,
financial and accounting models, market prediction models and to some extent sentimental analysis
which gives how consumers react to a given product. Its integration in the business world has led to
improved sales where

8
susmita dhital(HND/3rd Semester)
 Research Report 2019

Amazon records to over 60% sales from its recommendation systems. With the increase in online
customers from large companies, replacing human support with chatbots has been a great success
whereby these Chabots are replacing human workforce in providing automated and real-time support.

Several industries are turning to AI to do tasks that were earlier accomplished by people. Financial
services industry incorporates AI to process massive data, locate fraud by identifying unusual
operations, communicate online with customers, and perform several other essential functions. With
regards to facial reorganization, voice recognition and ML there are extremely some great beneficial
cases. New technologies are providing great benefits to enhance customer value propositions to drive
efficiency and effectiveness in the organization.

AI tools can bring essential benefits to the world of finance, which enable some tasks to be automated,
which give a boost for analytical capacity in comparison with traditional techniques that are now fast
becoming outdated. But these in spite of the life-changing benefits AI applications offer, they also
come with different limitations, which can make them not be suitable for performing some activities,
coupled with a range of risks that have to be adequately managed appropriately.

This thesis is organized in the following way. In Chapter 2, we introduce the term “AI”, its types and
its evolution over the years. Chapter 3 deals with the conceptual analysis of AI which studies the
opportunities and challenges of AI. In Chapters 4 and 5, we continue the research by discussing the
adoption of AI globally as well as in Finland. Chapter 6 is mainly based on the description of the
research process used for this thesis. Document analysis of qualitative analysis is used for this research.
It comprises of a list of electronic documents and publications that are used for the research. Chapter
7 analyzes the results of the study on the topic of AI in finance. This chapter is based on the study of
how AI is changing the financial service industry, the applications of AI in the financial sector and its
impact. Required recommendation is presented for the businesses that are planning to adopt AI in their
organizational function. Chapter 8 concludes this thesis and summarizes all that has been done in the
study.

9
susmita dhital(HND/3rd Semester)
 Research Report 2019

10
susmita dhital(HND/3rd Semester)