Asosiy kontentga oʻtish
proceeding

The Convergence of AI, Machine Learning, and Big Data in Financial Risk Management

Nodira OlimovaFergana State UniversityNurmukhammad SotvoldiyevBakhtiyor KholmatovFergana State UniversityShukhratjon AbdisamatovFergana State UniversityShakhlo NematovaFergana State UniversityShahbozkhuja KarimovFergana State University
2026en
Ochiq kirishOltin

Litsenziya: cc-by

ABI

Annotatsiya

The convergence of artificial intelligence, machine learning, and big data analytics in financial risk management is not merely a capability upgrade from logistic regression to gradient boosting-it represents a structural transformation of the risk intelligence pipeline from backward-looking actuarial modelling to real-time predictive risk sensing that processes orders of magnitude more data at higher velocity and granularity than regulatory frameworks were designed to govern. This paper proposes a Convergent AI-ML-Big Data Risk Intelligence Framework (CAMBRI) integrating a big data ingestion and real-time market signal processing engine, a multi-layer machine learning risk assessment module, and an explainable AI regulatory compliance and stress-testing gateway, evaluated across 58 financial institutions (banks$\mathbf{n} \boldsymbol{=} \mathbf{2 2}$, insurance$\mathbf{n} \boldsymbol{=} \mathbf{1 6}$, asset managers$\mathbf{n} \boldsymbol{=} \mathbf{1 4}$, fintech$\mathbf{n}=\mathbf{6})$processing 9,240,000 financial transaction records (6,468,000 training / 2,772,000 test, 70/30 chronological stratification) and 248,400 risk assessment cases across 5 risk categories (credit, market, operational, liquidity, systemic) over a 29-month evaluation horizon. The CAMBRI achieves a composite risk management effectiveness score of$\mathbf{r}=\mathbf{0. 8 9 7}$against expert-validated risk outcome benchmarks, outperforming ML-only risk analytics$(\mathbf{r}=\mathbf{0. 6 9 9}; +\mathbf{2 8. 3 \%}[(\mathbf{0. 8 9 7 - 0. 6 9 9}) / \mathbf{0. 6 9 9}])$, rule-based risk models$(\mathbf{r}=\mathbf{0. 5 9 8}; +\mathbf{5 0. 0} {\%}$[$(\mathbf{0. 8 9 7}-\mathbf{0. 5 9 8}) / \mathbf{0. 5 9 8}])$, and manual actuarial assessment (r = 0.484; +85.3% [(0.897-0.484)/0.484]). Deployment outcomes: +36.8% Value-at-Risk accuracy, +62.0% portfolio stress-test coverage, +54.0% fraud detection improvement, 92.4% regulatory compliance rate, -24.2% model computation time, and +38.2% real-time risk alerting speed. Sub-domain performance: Credit Risk AI$(\mathbf{r}=\mathbf{0. 9 3 8})$, Market Risk ML (0.936), Operational Risk Big Data (0.889), and Systemic Risk Integration (0.855). XAI compliance constraints degrade CAMBRI accuracy to 0.843 (-6.0% [(0.897-0.843)/0.897]), and heterogeneous data quality to 0.826 (-7.9% [(0.897-0.826)/0.897]), with deployed mitigation protocols for both conditions.

Hali tarjima qilinmagan

Identifikatorlar

Iqtiboslar va manbalar