Methodological aspects of neural network-based analysis of financial reporting
Amid the accelerating digitalization of the economy and the increasing complexity of the financial environment, there is a growing demand for more accurate, scalable, and intelligent tools for financial statement analysis. Traditional approaches based on deterministic models, financial ratio calculations, and expert interpretation are losing effectiveness in the face of large volumes of heterogeneous data and exhibit limited capacity to detect latent patterns or forecast risks. In this context, the application of artificial intelligence methods (particularly neural network models) has emerged as a promising avenue of both academic inquiry and practical innovation in the field of accounting. This article explores the methodological foundations of employing neural networks for the analysis of financial statements. The primary objective of the study is to systematize contemporary approaches to the construction, training, and interpretation of neural network architectures used for financial data analysis, while also identifying their advantages, limitations, and prospects for practical implementation in accounting and auditing. The object of the research is the financial statements of economic entities, considered as a source of data for analytical processing. The focus of the study is the application of neural network methods to the analysis of this information. The article provides a comprehensive review of recent international studies on the use of neural networks for forecasting financial indicators, diagnosing financial distress, and detecting fraud and anomalies in accounting records. Particular attention is paid to data preprocessing and standardization techniques, model architecture selection, and training procedures. Drawing on both international and Russian experiences, the article highlights the key benefits of neural networks, including enhanced classification accuracy, improved forecasting precision, broader data coverage, accelerated analysis, and the discovery of latent patterns inaccessible to traditional methods. At the same time, it emphasizes several challenges and constraints - from models' sensitivity to data quality and the need for large training datasets, to issues of result interpretability and the lack of unified methodological standards. The scientific novelty of this study lies in its integrated assessment of the potential of neural network technologies within the framework of modern financial analysis methodology. It also offers practical recommendations for integrating neural network tools into accounting, analytical, and audit practice. The article presents a systematized set of conclusions, a classification of neural network architectures, a synthesis of methodological approaches, and directions for further research. The author declares no conflicts of interests.
Keywords
financial statements,
neural networks,
financial analysis,
digital transformation,
artificial intelligence,
anomaly detection,
performance forecastingAuthors
| Popravko Inna V. | Voronezh State University | ipopravko@mail.ru |
Всего: 1
References
Kim A., Muhn M., Nikolaev V. Financial Statement Analysis with Large Language Models // arXiv preprint arXiv. 2024. 2407. 17866. doi: 10.48550/arXiv.2407.17866 (дата обращения: 28.05.2025).
Болотов Р.О., Суглобов А.Е. О применении нейронных сетей для оценки финансовой устойчивости компаний // Russian Journal of Management. 2020. Т. 8, № 1. С. 106-110. doi: 10.29039/2409-6024-2020-8-1-106-110 (дата обращения: 28.05.2025).
Huang H., Yuan Y. Automated Financial Statement Analysis Based on Deep Learning Neural Networks // Frontiers in Artificial Intelligence and Applications. 2025. № 391. Р. 847853. doi: 10.3233/FAIA241179 (дата обращения: 28.05.2025).
Coakley J.R., Brown C.E. Artificial Neural Networks in Accounting and Finance: Mod eling Issues // International Journal of Intelligent Systems in Accounting, Finance & Management. 2000. Vol. 9. Р. 119-144. doi: 10.1002/1099-1174(200009)9:3<119::AID- ISAF165>3.0.CO;2-9. URL: https://download.clib.psu.ac.th/datawebclib/e_resource/trial_da-tabase/WileyInterScienceCD/pdf/ISAF/ISAF_1.pdf (дата обращения: 28.05.2025).
Chen Y., Guo J., Huang J., Lin B. A novel method for financial distress prediction based on sparse neural networks with L1/2 regularization // International Journal of Machine Learning and Cybernetics. 2022. № 13. Р. 2089-2103. doi: 10.1007/s13042-022-01566-y (дата обращения: 28.05.2025).
Bi W., Xu B., Sun X., Wang Z., Shen H., Cheng X.Company-as-Tribe: Company Financial Risk Assessment on Tribe-Style Graph with Hierarchical Graph Neural Networks // arXiv preprint arXiv: 2023. 2301. 13492. doi: 10.48550/arXiv.2301.13492 (дата обращения: 28.05.2025).
Giantsidi S., Tarantola C. Deep Learning for Financial Forecasting: A Review of Recent Advancements // SSRN. 2025. doi: 10.2139/ssrn.5263710 (дата обращения: 28.05.2025).
Lipelis M. Bridging Financial Forecasting and Budget Management: New Approaches for Modern Enterprises // Scientific Journal of Bielsko-Biala School of Finance and Law. 2025. № 29 (1). doi: 10.19192/wsfip.sj1.2025.5 (дата обращения: 28.05.2025).
Artene A.E., Domil A.E. Neural Networks in Accounting: Bridging Financial Forecasting and Decision Support Systems // Electronics. 2025. № 14 (5). Р. 993. doi: 10.3390/electronics14050993 (дата обращения: 28.05.2025).
Rahayu D.V., Widuri R. Artificial Neural Network Methodology in Financial Statements Fraud: An Empirical Study in the Property and Real Estate Sector // Risk Governance & Control: Financial Markets & Institutions. 2025. № 15 (1). Р. 237-248. doi: 10.22495/rgcv15i1sip9 (дата обращения: 28.05.2025).
Muller R., Schreyer M., Sattarov T., Borth D. RESHAPE: Explaining Accounting Anomalies in Financial Statement Audits by enhancing SHapley Additive explanations // arXiv preprint arXiv. 2022. 2209. 09157. doi: 10.48550/arXiv.2209.09157 (дата обращения: 28.05.2025).
Bieganowski B., Slepaczuk R. Supervised Autoencoder MLP for Financial Time Series Forecasting // arXiv preprint arXiv. 2024. 2404. 01866. doi: 10.48550/arXiv.2404.01866 (дата обращения: 28.05.2025).
Li W., Liu X., Zhou S. Deep Learning Model Based Research on Anomaly Detection and Financial Fraud Identification in Corporate Financial Reporting Statements // Journal of Combinatorial Mathematics and Combinatorial Computing. 2024. Vol. 123. Р. 343-355. doi: 10.61091/jcmcc123-24.
Zhang Y., Yang X. Bridging Language Models and Financial Analysis // arXiv preprint arXiv. 2025. 2503. 22693. doi: 10.48550/arXiv.2503.22693.
Schreyer M., Hemati H., Borth D., Vasarhelyi M.A. Federated Continual Learning to Detect Accounting Anomalies in Financial Auditing. arXiv preprint arXiv: 20222210.15051. DOI: 10.48550/arXiv.2210.15051 (дата обращения: 28.05.2025).
Yang L., Li J., Dong R., Zhang Y., Smyth B. NumHTML: Numeric-Oriented Hierarchical Transformer Model for Multi-task Financial Forecasting // arXiv preprint arXiv. 2022. 2201. 01770. doi: 10.48550/arXiv.2201.01770 (дата обращения: 28.05.2025).
Elend L., Tideman S.A., Lopatta K., Kramer O. Earnings Prediction with Deep Learning // arXiv preprint arXiv. 2020. 2006. 03132. doi: 10.48550/arXiv.2006.03132 (дата обращения: 28.05.2025).
Wang C., Wang Y. Stock return prediction with multiple measures using neural network models // Financial Innovation. 2024. № 10. Р. 72. doi: 10.1186/s40854-023-00608-w (дата обращения: 28.05.2025).
Trinh V. Applicability of Deep Neural Networks in Business Decision Making and Market Prediction Investment // arXiv preprint arXiv. 2024. 2502. 00151. doi: 10.48550/arXiv.2502.00151 (дата обращения: 28.05.2025).
Scherrmann M., Elsas R. Earnings Prediction Using Recurrent Neural Networks // arXiv preprint arXiv. 2023. 2311. 10756. doi: 10.48550/arXiv.2311.10756 (дата обращения: 28.05.2025).
Durairaj M., Mohan B.H.K. A convolutional neural network based approach to financial time series prediction // Neural Computing and Applications. 2022. № 34. Р. 13319-13337. doi: 10.1007/s00521-022-07143-2 (дата обращения: 28.05.2025).
Golbayani P., Wang D., Florescu I. Application of Deep Neural Networks to Assess Corporate Credit Rating // arXiv preprint arXiv. 2020. 2003. 02334. doi: 10.48550/arXiv.2003.02334 (дата обращения: 28.05.2025).
Shabani M., Tran D.T., Magris M., Kanniainen J., Iosifidis A. Multi-head Temporal Attention-Augmented Bilinear Network for Financial Time Series Prediction // arXiv preprint arXiv. 2022. 2201. 05459. doi: 10.48550/arXiv.2201.05459 (дата обращения: 28.05.2025).
Cuervo R. Predictive AI for SME and Large Enterprise Financial Performance Management // arXiv preprint arXiv. 2023. 2311. 05840. doi: 10.48550/arXiv.2311.05840 (дата обращения: 28.05.2025).
Piven A. Analysis of Financial Reports in Companies Using Machine Learning // Financial Markets, Institutions and Risks. 2023. № 7 (4). Р. 135-154. doi: 10.61093/fmir.7(4).135-154.2023 (дата обращения: 28.05.2025).
Zim M.N.F. Harnessing Graph Neural Networks for Enhanced Fraud Detection in Financial Transactions // arXiv preprint arXiv. 2025. 2503. 22681. DOI: 10.48550/arXiv.2503.22681 (дата обращения: 28.05.2025).
Zouaghia Z., Kodia Z., Ben Said L. SMAPF-HNNA: A novel Stock Market Analysis and Prediction Framework using Hybrid Neural Network Architectures Across Major U.S. Indices // International Journal of Data Science and Analytics. 2025. doi: 10.1007/s41060-025-00811-1 (дата обращения: 28.05.2025).
Wang J., Zhang S., Xiao Y., Song R. A Review on Graph Neural Network Methods in Financial Applications // arXiv preprint arXiv. 2021. 2111. 15367. doi: 10.48550/arXiv.2rn.15367 (дата обращения: 28.05.2025).
