Volume 9, Issue 5, May – 2024 International Journal of Innovative Science and Research Technology

ISSN No:-2456-2165 https://doi.org/10.38124/ijisrt/IJISRT24MAY2125

An Overview of AI and Advanced Algorithmic

Applications in Financial Risk
Moussab El khair Ghoujdam*1; Rachid Chaabita2; Salwa idamia3; Oussama El khalfi4; Hicham El Alaoui5; Kamal Zehraoui6
1
Doctor in Economic Sciences, University Hassan II Casablanca, Maroc
2
Professor of Higher Education at the Faculty of Legal, Economic and Social Sciences - University Hassan II and Director of the
Laboratory Business Intelligence Governance of Organizations, Finance and Financial Crime. University Hassan II Casablanca,
Maroc
3
Doctor in Economic Sciences, University Hassan II Casablanca, Maroc
4
Research Professor, University Sidi Mohamed Ben Abdallah de Fes, FSJES, Maroc
5
Doctor in Economic Sciences, University Hassan II Casablanca, Maroc
6
Research Professor, University Hassan II Casablanca, Maroc

Abstarct:- This article delves into the transformative effects system led to a 150% increase in small loans granted by Baidu
of Artificial Intelligence (AI) and Machine Learning (ML) without an increase in credit losses within just two months.
on the realm of risk management. AI and ML technologies Machine learning is also applied to assess credit risks, optimize
have revolutionized risk assessment, mitigation, and investment strategies, and enhance overall risk management.
management across various sectors by offering advanced An examination of annual bank reports highlights the areas
analytical capabilities and automated decision-making where ML techniques have been implemented and the
processes. In the financial sector, for instance, these algorithms used [2]. Nevertheless, certain aspects of risk
technologies have facilitated improvements in loan decision management require further exploration and research to fully
processes, fraud detection, and compliance. Partnerships understand the potential and limitations of AI and ML
like ZestFinance and Baidu exemplify the successful technologies [3,4].
application of AI in enhancing loan decisions based on vast
data analysis. Despite the evident benefits, challenges such As new financial technologies and digital banking emerge,
as model-related risks, data availability and protection, and risk management practices are evolving accordingly. AI
the need for skilled personnel persist. This article aims to research and solutions are expected to make significant
provide a comprehensive overview of the current contributions to this evolution [5]. However, despite the
applications of AI and ML in risk management while numerous positive effects of AI, ML, and deep learning (DL),
identifying opportunities for further research and it is essential to consider the challenges and open questions.
development in this rapidly evolving field. These challenges include model-related risks ("black box"),
data availability, collection, and protection, transparency,
Keywords:- Artificial Intelligence (AI) ; Machine Learning ethics, and the need for skilled personnel to develop and
(ML); Risk Management; Credit Risk; Market risk; Operational implement new techniques (Financial Stability Board, 2017).
Risk.
The objective of this article is to provide an overview of
I. INTRODUCTION the current applications of AI and ML in risk management,
highlighting their advantages and identifying opportunities for
The integration of artificial intelligence (AI) and machine future research and development. Exploring the integration of
learning (ML) techniques into risk management is significantly AI and ML techniques in risk management is of major interest
transforming various sectors by offering advanced methods for for several reasons. Firstly, AI and ML have the potential to
assessing, mitigating, and managing risks. These technologies radically transform how risks are identified, assessed, and
enable precise data analysis and automated decision-making, managed, enabling faster, more accurate, and more
essential for navigating complex risk environments. comprehensive analyses than traditional methods. Secondly,
the adoption of AI and ML in risk management addresses the
In the financial sector, AI and ML are revolutionizing risk growing need to handle increasingly large and complex data
management by enhancing loan decision processes, detecting volumes. This leads us to pose the question: How can the
fraud, and improving compliance. For instance, ZestFinance integration of AI and ML techniques into risk management
leverages AI to analyze vast amounts of data for quick and improve the efficiency and accuracy of risk management
accurate loan decisions. In collaboration with Baidu, processes?
ZestFinance improved loan decisions in China by using data
points such as customers' search and purchase histories [1]. This

IJISRT24MAY2125 www.ijisrt.com 3039

Volume 9, Issue 5, May – 2024 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.38124/ijisrt/IJISRT24MAY2125

To answer this question, we began our discussion by

establishing a foundational understanding of risk management
and machine learning, setting the stage for a comprehensive
exploration of their integration. Subsequently, we delved into
the practical application of machine learning techniques within
the realm of risk management.

II. BACKGROUND

A. Risk Management
Managing risk in the banking sector is a significant
challenge for financial institutions, given their pivotal role in
economic systems. The main goal of risk management in banks
is to maintain financial stability while maximizing shareholder
returns. However, striving for profitability naturally leads to
greater exposure to various risks, which can threaten the
institution's long-term sustainability.

Banking institutions face numerous risks, including

interest rate risk, market risk, credit risk, off-balance sheet risk,
technological and operational risk, foreign exchange risk,
sovereign risk, liquidity risk, and default risk (Figure 1) [6]; [7].
Each type of risk has distinct characteristics and origins. For
instance, interest rate risk arises from market interest rate
fluctuations, while credit risk pertains to the potential default
by borrowers. Liquidity risk, which is treated separately from
other risks, manifests in two forms: asset liquidity risk and
funding liquidity risk. Asset liquidity risk occurs when a
transaction cannot be completed at current market prices, often
due to the transaction size relative to typical market lots.
Funding liquidity risk, also known as cash flow risk, refers to
the inability to meet cash flow obligations [8].

According to the Basel Committee on Banking

Supervision (BCBS), operational risk involves losses resulting
from internal processes, human or system failures, or external
events. This risk is a critical component of risk management
within banks, encompassing legal risks but excluding strategic
and reputational risks. Operational risk is inherent in all
banking products, activities, processes, and systems (Basel
Committee on Banking Supervision, 2011).

Fig 1: Taxonomy of Risks

Source: Leo, M., Sharma, S., & Maddulety, K. (2019)

IJISRT24MAY2125 www.ijisrt.com 3040

Volume 9, Issue 5, May – 2024 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.38124/ijisrt/IJISRT24MAY2125

To address this complexity, regulators have established Advancements in machine learning have paved the way
standards and guidelines to oversee bank risk management. The for the emergence of intelligent systems that rival human
Basel Accords, initiated in 1998 and updated since then, cognition. These systems are now ubiquitous in both our
provide a crucial regulatory framework for determining banks' professional and personal lives, influencing our interactions in
capital requirements based on the risks they face [9]; (Basel electronic markets in various ways. They contribute to
Committee on Banking Supervision, 2008). These accords improving decision-making in businesses, thereby enhancing
establish methods for calculating capital requirements for each productivity, employee engagement, and retention [12]. They
type of risk, enabling banks to measure and effectively manage also power personalized assistance systems tailored to
their risk exposure. Effective risk management also requires individual user preferences [13], as well as trading agents that
constant monitoring of market conditions and regular disrupt traditional financial markets [14].
assessment of risk portfolios. Banks must be able to identify
emerging trends and potential vulnerabilities, adjusting their Over the past few decades, the field of machine learning
strategies accordingly. This often involves the use of has witnessed significant advancements, particularly in the
sophisticated forecasting models and data analysis techniques development of sophisticated algorithms and effective
to anticipate market movements and assess risk scenarios. preprocessing techniques. One such major advancement has
Additionally, banks must also ensure adequate levels of been the transition from Artificial Neural Networks (ANNs) to
liquidity to meet fund withdrawal demands and potential deeper neural network architectures, characterized by enhanced
market shocks. Liquidity risk, manifested by a bank's inability learning capabilities, collectively referred to as Deep Learning
to honor its financial obligations, can have serious (DL) [15]. In specific environments, DL already demonstrates
consequences on its financial stability if not managed performance surpassing that of humans [16]; [17].
appropriately. Therefore, banks must develop robust liquidity
management policies and implement monitoring mechanisms In machine learning, three primary types are generally
to ensure effective risk management. recognized: supervised learning, unsupervised learning, and
reinforcement learning. While many applications in electronic
B. Machine Learning, markets utilize supervised learning, such as stock market
A sub-discipline of artificial intelligence, is a self-learning prediction, understanding customer perceptions, analyzing
process based on algorithms (Figure 2). It has the ability to customer needs, and product recommendation, other
analyze data and detect patterns without requiring explicit methodologies like unsupervised learning and reinforcement
programming for each task. For example, by exposing the learning are also being explored [18]; [19]; [20].
system to different types of data, it can learn and adapt, thereby
improving its performance over time [10]. This field of artificial  Supervised Learning: This approach involves training a
intelligence aims to develop models and techniques that enable computer program on labeled data to establish the
computers to learn from data, generalize knowledge, and make relationship between inputs and outputs. The process
decisions or formulate autonomous predictions [11]. requires manual labeling of data, making it particularly
useful when there is sufficient knowledge of the dataset.
Feature engineering, parameter tuning, and algorithm
selection must be performed by an expert. Supervised
learning algorithms are used to address regression and
classification tasks, optimizing parameters and minimizing
errors to predict continuous values or discrete classes from
input variables.
 Unsupervised Learning: Unlike supervised learning,
unsupervised learning operates on unlabeled data, removing
the need for manual labeling. It is mainly applied when there
is insufficient knowledge of the input data, aiming to group
the data into different patterns. Unsupervised learning
algorithms are employed for tasks like clustering,
dimensionality reduction, and anomaly detection.
 Semi-Supervised Learning: Combining aspects of both
supervised and unsupervised learning, semi-supervised
learning leverages a small amount of labeled data alongside
a larger pool of unlabeled data to enhance the accuracy and
Fig 2: The Correlation between Artificial Intelligence (AI), robustness of machine learning algorithms, reducing the
Machine Learning (ML), Deep Learning (DL), need for extensive manual labeling.
Neural Networks (NN), and Spiking Neural Networks (SNN).
Source: (Shinde &al,2018)

IJISRT24MAY2125 www.ijisrt.com 3041

Volume 9, Issue 5, May – 2024 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.38124/ijisrt/IJISRT24MAY2125

 Reinforcement Learning: This method depends on a trial- Galindo and Tamayo (2000) conducted a comparative
and-error process and a feedback mechanism to refine analysis of statistical classification and machine learning
previous states and actions, optimizing the developed techniques on credit portfolios, ranking the performance of over
function. In reinforcement learning, agents observe and 9000 models. They found that CART decision tree models
interact with an environment, receiving rewards or provided the best default estimates, with neural networks
punishments based on their actions, which helps in updating ranking second [33]. Hamori et al. (2018) compared the forecast
the machine learning model. This technique is especially accuracy and classification ability of various machine learning
beneficial for decision-making processes, such as deriving methods, concluding that boosting was superior [34]. Hybrid
optimal policy solutions. techniques and ensemble methods have also been explored for
credit scoring, where one technique is used for final predictions
III. APPLICATIONS OF MACHINE LEARNING after employing several others in the analysis [35]-[38]. For
AND AI IN RISK MANAGEMENT instance, an ensemble learning method using regularized
logistic regression, clustering, and bagging algorithms
A. Applications of machine learning and AI in credit risk outperformed many popular credit scoring models [39].
Credit risk refers to the potential economic loss from a
counterparty's failure to meet contractual obligations, such as B. Application of machine learning in market risk
timely payments of interest or principal, or an increased default Risk is quantified by the standard deviation of unexpected
risk during the transaction period. Traditionally, financial outcomes, often referred to as volatility. Value at Risk (VaR)
institutions used methods like linear regression, logit, and estimates the maximum loss over a target period that will not
probit models to assess credit risk [21]. However, there is a be exceeded with a specified level of confidence, capturing both
growing interest in employing artificial intelligence and underlying volatility and financial risk exposure. Accurately
machine learning techniques to enhance credit risk management forecasting market volatility is essential for risk management
practices, due to the limitations of traditional methods. AI and and asset pricing. Neural network models have been shown to
machine learning have proven to significantly improve credit improve volatility estimation methods. For example, Zhang et
risk management by effectively interpreting unstructured data. al. (2017) developed a model combining the Generalized
Autoregressive Conditional Heteroskedasticity (GARCH)
The use of AI and machine learning in credit risk model and the Extreme Learning Machine (ELM) algorithm to
modeling is not new but is expanding. In 1994, Altman and predict volatility. This model uses GELM-RBF to forecast time
colleagues conducted a comparative analysis of traditional series volatility and extrapolates these predictions to calculate
statistical methods and a neural network algorithm for VaR more accurately and efficiently. It employs a stochastic
predicting distress and bankruptcy, finding that a combined mapping method that is nonlinear and does not require Gaussian
approach improved accuracy [22]. The increasing complexity probability assumptions.
of credit risk assessment has led to the adoption of machine
learning, particularly in the context of credit default swaps Market risk also encompasses interest rate risk and
(CDS). Analyzing daily CDS data from January 2001 to exchange rate risk. Interest rate curves, depicting the
February 2014, non-parametric machine learning models, relationship between interest rates and debt maturity for a
including deep learning, were shown to outperform traditional specific borrower in a particular currency, are crucial in
models in forecast accuracy and practical coverage measures financial engineering and market risk management. The
[23]. "Gaussian Mixture Model" clustering method can be used to
create nonlinear models of parameter evolution and predict
In the domains of consumer and SME lending, vast interest rate curves, improving interest rate visualization.
amounts of data are increasingly utilized with machine learning Machine learning clustering techniques designed for stochastic
to enhance lending decisions. Companies like ZestFinance differential equations can help develop anticipatory VaR
operate in this area, demonstrating the effectiveness of machine models, crucial for managing risk during market regime
learning. For example, a technique based on decision trees and changes.
Support Vector Machines (SVM) resulted in significant cost
savings when applied to real lending data [24]. Similarly, Figini Liquidity risk is another significant concern that can be
et al. (2017) found that a multivariate outlier detection machine addressed with machine learning. Techniques like artificial
learning technique improved credit risk estimation for SME neural networks (ANNs) and genetic algorithms can be used to
loans [25]. SVM, a supervised learning algorithm used for measure liquidity risk, analyze key factors, and study the
classification, has been applied in various forms to design credit interconnections between these factors. ANNs can estimate
risk assessment and scoring models [26]. Studies have shown general risk trends and identify the most influential factors,
that models with a broader definition of credit risk are more while Bayesian networks can predict the probability of liquidity
accurate [27]-[32]. risk events.

IJISRT24MAY2125 www.ijisrt.com 3042

Volume 9, Issue 5, May – 2024 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.38124/ijisrt/IJISRT24MAY2125

Market risk, arising from investments, transactions, and network analysis can also be used to monitor employees and
overall financial market exposure, is another area where traders. Clustering and classification techniques can be
machine learning can be impactful. According to [46], machine employed to establish trader behavior profiles, where
learning can enhance market risk management at various stages, combining trading data, electronic and vocal communications
including data preparation, modeling, stress testing, and model records allows banks to observe emerging behavior patterns to
validation. Trading in financial markets carries the risk that the predict latent risks and detect connections between employees.
trading model may be incorrect, incomplete, or outdated, This also enables banks to generate and prioritize alerts based
known as model risk management. Machine learning is on types of suspicious activities and the level of involved risk.
particularly effective for stress testing market models to Ngai et al. (2011)[54] provide an excellent overview of the
identify emerging risks in trading behaviors. Woodall (2017) main AI techniques used for financial fraud detection, noting
discusses how machine learning is currently used in model that the main applied techniques are decision trees and neural
validation, such as Natixis's use of unsupervised learning to networks. For instance, five of the largest Nordic banks recently
discover new asset connection patterns. An exciting future joined forces to establish a common infrastructure for
application is the integration of reinforcement learning into combating money laundering, known as the Nordic KYC
market trading algorithms, allowing them to learn from market Utility. AI-based infrastructure will aid in complying with KYC
reactions and adapt their trading strategies accordingly. (Know Your Customer) regulations and requirements and
Additionally, [50] suggest combining neural networks and avoiding fines imposed by regulators. Similarly, HSBC is
decision tree techniques to provide real-time alerts to traders introducing AI technology, developed by data analytics firm
about changes in trading patterns. Quantexa, to monitor their anti-money laundering processes.
There are also practical efforts in fraud prevention. For
C. Application of Machine Learning in Operational Risk example, a joint venture of the Royal Bank of Scotland and
Machine learning is also applied in operational domains to Vocalink in the UK is creating a machine learning system to
enable risk mitigation, specifically in risk detection and/or analyze transactions from commercial clients, both large and
prevention. In terms of operational risk, aside from small, to identify and circumvent false invoices and potential
cybersecurity cases, machine learning primarily focuses on frauds. A study by Colladon and Remondi (2017) [55] using
issues related to fraud detection and suspicious transactions. real data from 33,000 transactions of an Italian factoring
Operational risk management involves the identification by company demonstrates the effectiveness of such analyses in
businesses of potential risks of direct or indirect financial loss fraud detection (see also Demetis, 2018)[56]. In terms of money
resulting from various operational failures [51]. These risks can laundering, criminals route money through various
be internal to institutions (e.g., inadequate or failing internal transactions, mixing them with legitimate transactions to
processes, human errors, or system failures) or can originate conceal the true source of funds. Funds usually originate from
from external events (such as fraud, vulnerable computer criminal or illegal activities and can be used in other illegal
systems, control failures, operational errors, neglected activities, including financing terrorist activities. There has
procedures, or natural disasters). The increase in the quantity, been extensive research on financial crime detection using
variety, and complexity of operational risk exposures, traditional statistical methods and, more recently, machine
particularly for financial firms, has led to the adoption of AI and learning techniques. Clustering algorithms identify customers
machine learning-based solutions [52]. AI can assist institutions with similar behaviors and can help find groups of people
at various stages of the risk management process, from working together to commit money laundering [57]. A major
identifying risk exposure to measuring, estimating, and challenge for banks, given the large volume of transactions per
evaluating their effects [53]. It can also help in choosing an day and the non-uniform nature of many transactions, is to be
appropriate risk mitigation strategy and finding instruments to able to sort through all transactions and identify those that are
transfer or negotiate risk. Thus, the use of AI techniques for suspicious. Financial institutions use anti-money laundering
operational risk management, which began with preventing systems to filter and classify transactions based on degrees of
external losses such as credit card frauds, now extends to new suspicion. Structured processes and intelligent systems are
areas involving analyzing vast collections of documents, needed to enable the detection of these money laundering
executing repetitive processes, and detecting money laundering transactions [58].
by analyzing large amounts of data.
IV. CONCLUSION
Financial fraud detection is another commonly referenced
use case of machine learning and AI in risk management. Banks Machine learning and artificial intelligence have
attempt to control financial fraud by evaluating the best ways to transformed financial risk management, covering credit,
protect their systems, data, and ultimately, their clients. The market, and operational risks. Our study has explored how these
ability of AI to automate processes can speed up routine tasks, technologies offer innovative and effective solutions for
minimize human errors, process unstructured data to filter identifying, assessing, and mitigating various types of risks.For
relevant content or negative news, and determine connections credit risk, AI and ML significantly improve the accuracy of
between individuals to assess at-risk clients and networks. This prediction models, surpassing traditional techniques like linear

IJISRT24MAY2125 www.ijisrt.com 3043

Volume 9, Issue 5, May – 2024 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.38124/ijisrt/IJISRT24MAY2125

and logistic regressions. Financial institutions can thus better [6]. Saunders, Anthony, Marcia Millon Cornett, and Patricia
assess the probability of borrower default and optimize lending Anne McGraw. 2006. Financial Institutions
decisions. For example, techniques like neural networks and Management: A Risk Management Approach. New
decision trees have shown promising results in reducing losses York: McGraw-Hill.
and managing loan portfolios. Regarding market risk, AI is used [7]. Apostolik, Richard, Christopher Donohue, Peter Went,
to enhance volatility forecasting and value-at-risk (VaR) and Global Association of Risk Professionals. 2009.
assessment. Models based on deep learning and non-parametric Foundations of Banking Risk: An Overview of Banking,
methods often outperform traditional models in terms of Banking Risks, and Risk-Based Banking Regulation.
accuracy and efficiency. These advancements enable New York: John Wiley.
institutions to better manage their exposure to interest rate [8]. Jorion, Philippe. 2007. Value at Risk: The New
fluctuations, currency risks, and stock price changes, while Benchmark for Managing Financial Risk. New York:
meeting increasingly complex regulatory requirements.Lastly, McGraw-Hill
in operational risk management, AI and ML focus on fraud [9]. Hull, John. 2012. Risk Management and Financial
detection and prevention of suspicious transactions. Clustering Institutions. New York: John Wiley and Sons, vol. 733.
and classification techniques help identify abnormal behaviors [10]. Bishop, C. M. (2006). Pattern recognition and machine
and fraudulent activities, while neural networks and logistic learning. Springer google schola, 2, 1122-1128.
regression algorithms are used to analyze large amounts of data [11]. Hastie, T., Tibshirani, R., Friedman, J. H., & Friedman,
and detect potential risks. These tools are also applied in J. H. (2009). The elements of statistical learning: data
combating money laundering and cybercrime, thereby mining, inference, and prediction (Vol. 2, pp. 1-758).
strengthening the security and compliance of financial New York: springer.
institutions. [12]. Shrestha, U., Alsadoon, A., Prasad, P. W. C., Al Aloussi,
S., & Alsadoon, O. H. (2021). Supervised machine
Despite these advancements, our research has also learning for early predicting the sepsis patient: modified
highlighted certain limitations. Integrating AI and ML into risk mean imputation and modified chi-square feature
management requires significant investments in infrastructure selection. Multimedia Tools and Applications, 80(13),
and training. Moreover, ML models can introduce algorithmic 20477-20500.
biases and modeling errors, necessitating robust governance [13]. Al-Dury, N., Ravn-Fischer, A., Hollenberg, J.,
and oversight frameworks. Transparency and interpretability of Israelsson, J., Nordberg, P., Strömsöe, A., ... &
models remain major challenges that need to be addressed to Rawshani, A. (2020). Identifying the relative
ensure effective and ethical adoption of these technologies. importance of predictors of survival in out of hospital
cardiac arrest: a machine learning study. Scandinavian
REFERENCES journal of trauma, resuscitation and emergency
medicine, 28, 1-8.
[1]. Technology Review. (2017). "An AI-fueled credit [14]. Balaji, A. J., Balaji, G. T., Dinesh, M. S., Nair, B. B., &
formula might help you get a loan." Retrieved from Ram, D. H. (2018, December). A machine learning
[https://www.technologyreview.com/s/603604/an-ai- based approach to crack detection in asphalt pavements.
fueled-credit-formula-might-help-you-get-a- In 2018 15th IEEE India Council International
loan/](https://www.technologyreview.com/s/603604/an Conference (INDICON) (pp. 1-4). IEEE.
-ai-fueled-credit-formula-might-help-you-get-a-loan/). [15]. Goodfellow, I., Bengio, Y., & Courville, A. (2016).
[2]. Westengen, O.T., Haug, R., Guthiga, P., and Macharia, Deep learning. The MIT Press
E. (2019). "Global climate risk management: The role of [16]. Madani, A., Arnaout, R., Mofrad, M., & Arnaout, R.
agribusiness." *Climate Risk Management*, 23, 18-28. (2018). Fast and accurate view classification of
[3]. Taneja, S., Awasthi, A., and Arora, P. (2020). "A echocardiograms using deep learning. Npj Digital
heuristic machine learning approach for agriculture Medicine, 1(1)
supply chain risk assessment." *Computers and [17]. Silver, D., Hubert, T., Schrittwieser, J., Antonoglou, I.,
Electronics in Agriculture*, 168, 105130. Lai, M., Guez, A., Lanctot, M., Sifre, L., Kumaran, D.,
[4]. Zhong, R.Y., and Zhou, Q. (2020). "The application of [18]. Brynjolfsson, E., & McAfee, A. (2017). The business of
artificial intelligence in smart agriculture: A survey." artificial intelligence. Harvard Business Review, 1–20
*IEEE Transactions on Industrial Informatics*, 16(4), [19]. Ramaswamy, S., & DeClerck, N. (2018). Customer
2978-2987. perception analysis using deep learning and NLP.
[5]. Kolanovic, M., and R. T. Krishnamachari, 2017, “Big Procedia Computer Science, 140, 170–178
Data and AI Strategies, Machine Learning and [20]. Kühl, N., Mühlthaler, M., & Goutier, M. (2020).
Alternative Data Approach to Investing”, Technical Supporting customeroriented marketing with artificial
Report, JP Morgan. intelligence:Automatically quantifying customer needs
from social media. Electronic Markets, 30(2), 351–367

IJISRT24MAY2125 www.ijisrt.com 3044

Volume 9, Issue 5, May – 2024 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.38124/ijisrt/IJISRT24MAY2125

[21]. Altman, E. I. (1968). Financial ratios, discriminant [34]. Hamori, Shigeyuki, Minami Kawai, Takahiro Kume,
analysis and the prediction of corporate bankruptcy. The Yuji Murakami, and Chikara Watanabe. 2018.
Journal of Finance, 23(4), 589–609. Ensemble Learning or Deep Learning? Application to
[22]. Altman, E. I. (1968). Financial ratios, discriminant Default Risk Analysis. Journal of Risk and Financial
analysis and the prediction of corporate bankruptcy. The Management 11: 12.
Journal of Finance, 23(4), 589–609. [35]. Bastos, João A. 2014. Ensemble Predictions of
[23]. Son, Y., Byun, H., & Lee, J. (2016). Nonparametric Recovery Rates. Journal of Financial Services Research
machine learning models for predicting the credit default 46: 177–93
swaps: An empirical study. Expert Systems with [36]. Hamori, Shigeyuki, Minami Kawai, Takahiro Kume,
Applications, 58, 210–220. Yuji Murakami, and Chikara Watanabe. 2018.
[24]. Khandani, Amir E., Adlar J. Kim, and Andrew W. Lo. Ensemble Learning or Deep Learning? Application to
2010. Consumer credit-risk models via machine- Default Risk Analysis. Journal of Risk and Financial
learning algorithms. Journal of Banking & Finance 34: Management 11: 12.
2767–87. [37]. Raei, Reza, Mahdi Saeidi Kousha, Saeid Fallahpour, and
[25]. Figini, S., Bonelli, F., & Giovannini, E. (2017). Mohammad Fadaeinejad. 2016. A Hybrid Model for
Solvency prediction for small and medium enterprises in Estimating the Probability of Default of Corporate
banking. Decision Support Systems, 102, 91–97. Customers. Iranian Journal of Management Studies 9:
[26]. Ray, Sunil. 2015. Understanding Support Vector 651–73.
Machine Algorithm from Examples (along with Code). [38]. Chen, Ning, Bernardete Ribeiro, and An Chen. 2016.
Available online: Financial Credit Risk Assessment: A Recent Review.
https://www.analyticsvidhya.com/blog/2017/09/underst Artificial Intelligence Review 45: 1–23.
aing-support-vector-machine-example- code [39]. Wang, Hong, Qingsong Xu, and Lifeng Zhou. 2015.
[27]. Bellotti, Tony, and Jonathan Crook. 2009. Support Large Unbalanced Credit Scoring Using Lasso-Logistic
Vector Machines for Credit Scoring and Discovery of Regression Ensemble. PLoS ONE 10: e0117844.
Significant Features. Expert Systems with Applications. [40]. Jorion, Philippe. 2007. Value at Risk: The New
[28]. Cao, Jie, Hongke Lu, Weiwei Wang, and Jian Wang. Benchmark for Managing Financial Risk. New York:
2013. A Loan Default Discrimination Model Using McGraw-Hill.
Cost-Sensitive Support Vector Machine Improved by [41]. Monfared, Soheil Almasi, and David Enke. 2014.
PSO. Information Technology and Management 14: Volatility Forecasting Using a Hybrid GJR-GARCH
193–204. Neural Network Model. Procedia Computer Science 36:
[29]. Van Gestel, Ir Tony, Bart Baesens, Ir Joao Garcia, and 246–53.
Peter Van Dijcke. 2003. A support vector machine [42]. Zhang, Wenhao. 2017. Machine Learning Approaches
approach to credit scoring. In Forum Financier—Revue to Predicting Company Bankruptcy. Journal of Financial
Bancaire Et Financiaire Bank En Financiewezen. Risk Management 6: 364–74.
Bruxelles: Larcier, pp. 73–82. Available online: [43]. Kanevski, Mikhail F., and Vadim Timonin. 2010.
http://citeseerx.ist.psu.edu/viewdoc/download?doi=10. Machine learning analysis and modeling of interest rate
1.1.93.6492&rep= rep1&type=pdf curves. Paper presented at the 18th European
[30]. Huang, Cheng Lung, Mu Chen Chen, and Chieh Jen Symposium on Artificial Neural Networks ESANN,
Wang. 2007. Credit Scoring with a Data Mining Bruges, Belgium, April 28–30; Available online:
Approach Based on Support Vector Machines. Expert https://www.elen.ucl.ac.be/Proceedings/esann/esannpdf
Systems with Applications 33: 847–56. /es2010-17.pdf.
[31]. Lai, Kin Keung, Lean Yu, Ligang Zhou, and Shouyang [44]. Mahdavi-Damghani, Babak, and Stephen Roberts. 2017.
Wang. 2006. Credit risk evaluation with least square A Proposed Risk Modeling Shift from the Approach of
support vector machine. In International Conference on Stochastic Differential Equation towards Machine
Rough Sets and Knowledge Technology. Learning Clustering: Illustration with the Concepts of
Berlin/Heidelberg: Springer, pp. 490–95. Anticipative and Responsible VaR. SSRN Electronic
[32]. Harris, Terry. 2013. Quantitative credit risk assessment Journal.
using support vector machines: Broad versus Narrow [45]. Tavana, Madjid, Amir Reza Abtahi, Debora Di Caprio,
default definitions. Expert Systems with Applications and Maryam Poortarigh. 2018. An Artificial Neural
40: 4404–13. Network and Bayesian Network Model for Liquidity
[33]. Galindo, Jorge, and Pablo Tamayo. 2000. Credit Risk Risk Assessment in Banking. Neurocomputing 275:
Assessment Using Statistical and Machine Learning: 2525–54.
Basic Methodology and Risk Modeling Applications.
Computational Economics 15: 107–43.

IJISRT24MAY2125 www.ijisrt.com 3045

Volume 9, Issue 5, May – 2024 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.38124/ijisrt/IJISRT24MAY2125

[46]. Kumar, P. P. (2018). Machine learning for model

development in market risk. GARP Institute. Available
at:https://www.garp.org/#!/riskintelligence/all/all/a1Z1
W000003fM0yUAE?utm_medium=social&utm_source
=face-
book&utm_content=org_whitepaper&utm_term=machi
nelearning&utm_campaign=sm_riskintelligence.
[47]. Financial Stability Board. (2017). Artificial intelligence
and machine learning in financial services. Available at:
http://www.fsb.org/wp-content/uploads/ P011117.pdf.
[48]. Woodall, L. (2017). Model risk managers eye benefits
of machine learning. Risk.net. Available at:
https://www.risk.net/risk-management/4646956/model-
risk-man- agers-eye-benefits-of-machine-learning. Last
[49]. Hendricks, D., & Wilcox, D. (2014). A reinforcement
learning extension to the Almgren-Chriss framework for
optimal trade execution. In IEEE Conference on
Computational Intelligence for Financial Engineering &
Economics (CIFEr) (pp. 457–464).
[50]. Chandrinos, S. K., Sakkas, G., & Lagaros, N. D. (2018).
AIRMS: A risk man- agement tool using machine
learning. Expert Systems with Applications, 105, 34–48.
[51]. Moosa, I. A. (2007). Operational risk management. New
York: Palgrave Macmillan.
[52]. Choi, T. M., Chan, H. K., & Yue, X. (2017). Recent
development in big data analytics for business
operations and risk management. IEEE Transactions on
Cybernetics, 47(1), 81–92.
[53]. Sanford, A., & Moosa, I. (2015). Operational risk
modelling and organizational learning in structured
finance operations: A Bayesian network approach.
Journal of the Operational Research Society, 66(1), 86–
115.
[54]. Ngai, E. W., Hu, Y., Wong, Y. H., Chen, Y., & Sun, X.
(2011). The application of data mining techniques in
financial fraud detection: A classification framework
and an academic review of literature. Decision Support
Systems, 50(3), 559–569.
[55]. Colladon, A. F., & Remondi, E. (2017). Using social
network analysis to prevent money laundering. Expert
Systems with Applications, 67, 49–58.
[56]. Demetis, D. S. (2018). Fighting money laundering with
technology: A case study of Bank X in the UK. Decision
Support Systems, 105, 96–107.
[57]. Sudjianto, Agus, Sheela Nair, Ming Yuan, Aijun Zhang,
Daniel Kern, and Fernando Cela-Diá z. 2010. Statistical
Methods for Fighting Financial Crimes. Technometrics
52: 5–19.
[58]. Kannan, Somasundaram, and K. Somasundaram. 2017.
Autoregressive-Based Outlier Algorithm to Detect
Money Laundering Activities. Journal of Money
Laundering Control 20: 190–202.

IJISRT24MAY2125 www.ijisrt.com 3046