ACADEMIC RESEARCH IN MODERN SCIENCE
International scientific-online conference
134
AUTOMATING DECISION-MAKING PROCESSES IN BANKING
FINANCIAL INFORMATION SYSTEMS USING ARTIFICIAL
INTELLIGENCE AND MACHINE LEARNING
Akbarova Marguba Khamidovna
Associate professor of the Department of "System and Application
Programming" of the Tashkent University of Information Technologies named
after Muhammad al-Khwarizmi.
Email: margubaakbarova66@gmail.com
Sharipov Bahodir Akilovich
Senior lecturer of the Department of "Systematic and Applied Programming" of
Tashkent University of Information Technologies named after Muhammad al-
Khwarizmi, Uzbekistan. Email: baxodiroqilovich1961@gmail.com
Djangazova Kumriniso Abdulvahobovna
Assistant of the Department of "Systematic and Applied Programming" of
Tashkent University of Information Technologies named after Muhammad al-
Khwarizmi, Uzbekistan.
Email: qumriniso17@gmail.com
Nurdullaev Alisher Niyatilla ugli
Assistant of the Department of "Systematic and Applied Programming" of
Tashkent University of Information Technologies named after Muhammad al-
Khwarizmi, Uzbekistan.
Email: Alishernurdullaev@gmail.com
Jurakulov Nodirbek Sobirovich
Assistant of rhe Department of Department of fundamentals of informaticsof
Tashkent university of informarion Technologies named after Muhammad al-
Kharizmi, Uzbekistan
Email: nodirbekj062@gmail.com
https://doi.org/10.5281/zenodo.15575044
Аннотация:
В данной работе рассматриваются перспективы
автоматизации
процессов
принятия
решений
в
финансовых
информационных системах банков с использованием искусственного
интеллекта и методов машинного обучения. Основное внимание
уделяется анализу существующих алгоритмов, их применению для оценки
кредитоспособности клиентов, управления рисками и выявления
мошеннических операций. Обсуждаются преимущества внедрения
интеллектуальных технологий, такие как повышение точности прогнозов,
снижение операционных издержек и ускорение обработки данных. Также
рассматриваются вызовы и потенциальные риски, связанные с
использованием ИИ в банковском секторе, включая вопросы этики,
безопасности данных и прозрачности алгоритмов.
ACADEMIC RESEARCH IN MODERN SCIENCE
International scientific-online conference
135
Ключевые слова:
искусственный интеллект, машинное обучение,
банковские информационные системы, автоматизация
решений,
финансовые технологии, управление рисками, кредитный скоринг, защита
данных, цифровая трансформация.
Abstract:
This paper analyzes the opportunities and advantages of
automating decision-making processes in banking financial information systems
using Artificial Intelligence and Machine Learning. AI and ML technologies have
been successfully applied in banking for risk management, credit scoring, fraud
detection, and optimizing investment strategies. By automating decision-making
with AI/ML, banks can achieve faster and more reliable results, while reducing
operational costs. The paper also explores system architecture, integration, and
the impact of these technologies on security concerns.
Keywords:
artificial intelligence, machine learning, decision-making,
automation, banking information system, credit scoring, fraud detection,
investment strategy, risk management, security
Login
Financial information systems today's of banking activities per day to the
center turned from lending pull investment up to the management was decision
acceptance to do processes done increases. Digital transformation, transaction
size sharp growth and customer requirements diversity - to banks decision
acceptance to do processes automation mandatory Artificial intellect and car
study technologies to banks not only real - time in mode information analysis to
do, maybe predictive from models using risk reduce and service quality improve
opportunity This is giving. article automation theoretical the basics of banking
decision acceptance to do in the process solutions integration to do methods and
also, this of the process economic and security aspects analysis does. Financial
information systems of banking operations heart They are the customers'
account -books management, credit and deposit operations management, money
flow control to do such as directly financial security and stability providing
functions does. With this together, digital transformation under the
circumstances transactions number and their complexity noticeable at the level
increased. In such circumstances decision acceptance to do processes
automation to banks following priority advantages gives:
Operation efficiency
– human factor reduction and processes acceleration
through service show speed increase
Accuracy and consistency
– standardized algorithmic approach with
decision quality stable at the level hold to stand
ACADEMIC RESEARCH IN MODERN SCIENCE
International scientific-online conference
136
Risks management
– big in size real time data in mode analysis to do
through credit and cyber -threat risks in advance to determine.
To regulators compliance
– decision acceptance to do processes
automated protocols based on conducted due to audit and accountable
requirements easy to do.
this purpose, banks financial information to systems artificial intelligence
and car learning components integration to do today's both practical and
theoretical at the level current research to the topic has become.
Home part
1. Artificial intellect and car study Theory:
Artificial to intelligent
banking systems in advance unwritten rules based on study, decision acceptance
to do and independent conclusion release ability gives. The machine study and
enlarged transaction information on forecast creative models basis organization
With the help of supervised learning credit scoring and customer scores clear
designation possible, unsupervised approaches through and anomalous
transactions fraud detection and customer segmentation done increase It gets
easier. Deep deep learning networks and complicated time rows and graphic
structured network information analysis to do possible. In a banking
environment this approaches together used, decided quality noticeable at the
level increases. Artificial the term intelligence computer to systems human to
the mind typical analysis, decision acceptance to do and study ability provider
technologies means. The machine learning is the study of SI concepts one
syllable from the data independent conclusion release and forecasting done
increases. -In the field of financial decision making the following MO approaches
wide used:
- Supervised Supervised Learning:
in advance designated using a sample
(label) the model to teach method. Credit used in scoring (logistic regression,
random forest), fraud detection ( gradient boosting, support vector machine) .
Model evaluation for such as accuracy, ROC - AUC, F1score indicators is used.
- Uncontrolled learning (Unsupervised Learning ):
in the information
hidden discovers structures. Clustering (K means, DBSCAN) credit portfolio
segmentation, anomalous detection (isolation forest, autoencoder) fraud in
determining service does.
- Deep Deep Learning:
one how many neuron layers using complicated
patterns learns. LSTM/ RNN time rows through market prices forecasting, GNN
(Graph Neural Networks ) between banks financial networks in analysis,
convolutional networks and documents automatic again at work is used.
ACADEMIC RESEARCH IN MODERN SCIENCE
International scientific-online conference
137
- Reinforcement Reinforcement Learning:
agent - environment
interaction optimal decision based on acceptance to do policy learns Dynamic
asset allocation and in algotrading, like Q - learning, deep Qnetwork methods
implementation is being done.
MO models working on the way out information data quality, errors
robustness, model consistency (stability), and understandable separate from
interpretability attention focus necessary.
1.Decision acceptance processes automation Methods:
In banks
decision acceptance to do three from the stage consists of: information
collection, analysis and decision SI /MO integration for first of all transactions,
customer profiles and market indicators continuous meeting necessary. Then
real -time information flow on predictive models implementation for example,
credit scoring gradient boosting algorithm for or implement an autoencoder for
fraud detection Finally, the model results mainly automatic decisions issuer
module ( for example, " debt" "to be given / not to be given " or " transaction"
stop / play "to be done ") and human control between balance This is created.
the process automation to banks operational expenses reduce and decision
acceptance to do speed to increase help gives.
2.System architecture and integration:
SI/MO models to practice current
to grow bank information for system many layered to architecture has to be
necessary. First of all, the data infrastructure – ETL processes through
information to the warehouse gathered and cleaned structured data model
training for Using containerization (Docker, Kubernetes) every as a model
microservice to work downloaded via API main system with integration Real
- time information Stream via Kafka or Flink directly - right to the model
transfer, result and RESTful interface or gRPC through customer to the services
to deliver practices current Also, MLOps process model versioning, monitoring
(Prometheus, Grafana) and through automated CI/CD system permanent quality
with performance is provided.
AI/ML components financial information to systems following architecture
principles based on integration is done:
1.Microservices and containerization:
every an ML model or service
using Docker / Kubernetes separately in the container to work is lowered;
2.API -first approach:
RESTful or gRPC interfaces through financial
information and forecast to the services connection;
ACADEMIC RESEARCH IN MODERN SCIENCE
International scientific-online conference
138
3.Data pipeline and MLOps:
Real time via Kafka or Flink information
currents Model training using transfer, Airflow or Kubeflow and update
automation;
4.Monitoring and logging:
with Prometheus/Grafana model work activity,
latency and accuracy drift monitoring; SIEM system through dangerous surveys
and anomalies to determine;
5.CI/CD:
Like GitLab CI, Jenkins tools using code and model versions
control, test transfer and fast maturity
Conclusion
Modern banks information in systems artificial intellect and car study
technologies application banking services automation, processes acceleration
and efficiency increase opportunity Today 's on the day banks decision
acceptance to do processes only financial services presented in the process of
not, maybe risks management, fraud determination and to customers service
also important in showing role plays. Artificial intellect and car study using
banks their own operational processes simplify and reduce costs reduces,
service show quality increases and security to strengthen achieves. From this In
addition, AI/ML systems decision acceptance to do in the process transparency
increase and in the system mistakes to minimize help However, such systems
current to be with related some problems For example, the data confidentiality,
model transparency and boot to the standards compliance to do such as issues
about still complete solutions working Bias ( reversal ) and of information to the
quality relatively caution with approach It is also necessary. systems users
wrong information or decisions based on to manage take absence for privacy
and security provision according to strict regulator requirements current to be
need. In the future artificial intellect and car study technologies in banks further
wider application is expected. Especially, Explainable AI technology decision
acceptance to do processes further understandable and reliable in doing help
Federated learning technology provides and to banks customers secret
information
Literature list:
1.
Kumar, S., & Verma, A. (2023). "Artificial Intelligence and Machine
Learning in Banking: Transforming Decision-Making Processes." Journal of
Banking Technology, 34(2), 157-169.
2.
Chen, X., & Yang, L. (2023). "Machine Learning Models for Credit Scoring: A
Comprehensive Review." Financial Technology and Innovations, 18(4), 45-58.
ACADEMIC RESEARCH IN MODERN SCIENCE
International scientific-online conference
139
3.
Singh, R., & Patel, S. (2023). "Fraud Detection in Banking Systems Using
Artificial Intelligence: A Comparative Study." Journal of Financial Security, 12(1),
77-91.
4.
Lopez, J., & Zhang, M. (2022). "AI-Driven Decision Making in Financial
Institutions: Opportunities and Challenges." Journal of Financial Innovation,
9(3), 123-134.
5.
Gupta, N., & Sharma, A. (2022). "Integrating Machine Learning Algorithms
in Financial Risk Management." Journal of Financial Engineering, 14(6), 230-
242.
6.
Buchanan, M., & Miller, C. (2023). "AI and Machine Learning for Banking:
Applications and Ethical Considerations." Banking Review and Innovation,
28(5), 200-212.
7.
Hassan, H., & Ali, N. (2023). "Exploring Blockchain and AI in Modern
Banking Systems." Journal of Digital Finance, 15(2), 111-122.
8.
Patel, K., & Gupta, V. (2023). "Advances in AI-Based Automated Decision
Systems in Banks." Financial Technology Journal, 8(4), 115-128.
