Professional Documents
Culture Documents
Literature Review
Literature Review
Literature Review
Introduction
The revolution of information and communication technology has changed the
lives and operations of individuals and organizations respectively. The global
spread of the Internet and its rapid use over the years, have contributed greatly to
the big change of e-commerce in the global business environment. So, business
transactions between trading partners continue to be offered on e-commerce
platforms. Electronic payment solutions have also been developed. The importance
of this development in the global business environment has challenged most
organizations to automatically switch from traditional paper money transactions to
electronic forms that provide more convenient, fast and secured process of making
payments among individuals and organizations. With the introduction of electronic
payment systems, the annual rate of global non-cash transactions facilitated by
electronic and mobile payments (m-payments) have increased significantly in the
past years. (In Banking a Mini-Review, February 2020)
The focal point of business in many enterprises is productivity and stability and the
financial business is the same in such manner. This stability is undeniably based on
proper decision making and artificial intelligence is incomparable in this cycle as it
removes the probability of human error from decision making process. Reasonable
preparation and dynamic planning without a doubt will lead to a successful
development. As the financial business begins to understand this point there is a
need to automate decision making processes in order to reduce the margin of error.
Deep learning is a high-level approach of machine learning based on artificial
neural network algorithms as a promising branch of artificial intelligence, recently
has drawn in huge interest. (Huang et al. Frontiers 2020)
E-PAYMENT SYSTEM
In the last twenty years, electronic payment system (EPS) has intrigued numerous
analysts and data framework experts due to their significant demonstration in
future electronic commerce. Thus, numerous profound and researches showed up
with various methodologies on electronic payment definitions and were basically
concentrated by gathering of information of academic reports from various
foundations like accounting and financial fields, business innovation to those in
data frameworks. For example, one specialist characterizes e-payment framework
as a type of commercial responsibility that includes the purchaser and the vender
worked with through the utilization of electronic correspondences. Likewise,
considers e-payments to be a type of connection between organizations and
individuals supported by banks and in-switch houses that empowers commercial
exchange electronically.
In another point of view, demonstrated e-payment system as any form of fund
transfer via the internet. Similarly, an electronic payment system refers to an
electronic purchase of products and services through online platforms or in
supermarkets and shopping malls. Moreover, electronic payments also defined as
payment by electronic transfer of credit card details, direct credit or other
electronic means other than payment by cheque and cash (Mohammad Auwal, Siti
Zabedah, Aidi Ahmi 2015).
Techniques
Expert Systems
Some reports focused specifically on expert systems, looked at as typical artificial
intelligence. An expert knowledge is documented in a database called knowledge
base. This base is connected to an inference base that allows answering back user
queries. Depending on the system, the user uses a port that could be more or less
detailed and questions the system to give suggestions based on the expert's
experience uploaded in the knowledge base. (Carlo Milana, Arvind Ashta 2021)
Machine learning
Machine Learning is characterized by mathematics, probability theory, and
statistics that consists of monitored readings, unendorsed readings, and enhanced
readings. Monitored readings uses tagged data and feedback if the outcome is
correct. In contrast, unendorsed readings must find its structure or patterns from
untagged data, and the system can learn on its own. Enhanced readings allow the
framework to learn for an online data, and credit assignment is based on positive
and negative messages. Enhanced readings could be beneficial in trading securities
that require a constant rebalancing of portfolios. (Carlo Milana, Arvind Ashta
2021)
Algorithms
Algorithms are a group of commands and orders used in mathematics or computer
science to identify and fix an issue. Thus, beginning from an input, resulting with
an output. Normally, an issue is discovered, a set of rules are designed to solve it,
and the solution is examined over a wide range of examples. There are thousands
of algorithms and different ways of classifying them. Even within the research on
finance and financial markets, there are many kinds of algorithms being used. One
example of such an algorithm is Classification and Regression Tree (CART)
analysis, an induction model that discovers rules from incomplete data. (Carlo
Milana, Arvind Ashta 2021)
Neural Networks
Neural networks (NNs) connect inputs with outputs but with hidden intermediate
layers of neurons in between. Each unit in each layer is connected to other units
and makes only one calculation. Each of these connections is given weights that
may be fixed or may evolve, and some networks even allow connections to be
dropped and new ones to be generated. (Carlo Milana, Arvind Ashta 2021)
Decision making
Scientists complied that all organizations, particularly those in competing
enterprises, should take on artificial intelligence into their firms so they could have
to compete with other firms which are already artificial intelligence adopted, so
they will not face failure and enhance their decision-making process. For instance,
artificial intelligence will permit the business to follow a brand picture shift
utilizing proof from the web and online entertainment. It could likewise find proof
of the reasons for brand shift and loss of portion of the overall industry.
Temporarily, human-made intelligence based independent direction is probably
going to work better where choices are unmistakable. More broad choices will in
any case require a human component. On other hand the estimates are dependable,
rules can be contrived to permit the program to decide or ideas to help the chiefs'
choices by permitting joining of various guidelines into one meta-rule and
recognizing which rule is applied. (Carlo Milana, Arvind Ashta 2021)
Bankruptcy
Bankruptcy forecasting is an important area of economic research because it is one
of interests to creditors, shareholder and governors. The most referred study in this
area has been the development of z-scores by Altman (1968) using characteristic
analysis. However, financial shareholder is interested bankruptcy predictions using
machine learning to enhance predictive ability. Initial research found that
transmitted algorithms were not as good as linear characteristic analysis in
detecting bankruptcy, despite of this findings still researchers confirms that
machine learning models scoring very good results in bankruptcy prediction.
(Carlo Milana, Arvind Ashta 2021)
References
Artificial intelligence techniques in finance and financial markets
May 2021
https://www.researchgate.net/publication/351465873
Artificial Intelligence: In Banking A Mini-Review
Feb 2020
https://www.researchgate.net/publication/339299131
Adoption of e-Payment Systems: A Review of Literature
October 2015
https://www.researchgate.net/publication/303329794
Deep learning in finance and banking: A literature review and classification
2020 http://creativecommons.org/licenses/by/4.0/