International Journal of Neutrosophic Science IJNS 2690-6805 2692-6148 10.54216/IJNS https://www.americaspg.com/journals/show/4000 2020 2020 Department of Corporate Finance and Corporate Governance, Financial University under the Government of the Russian Federation, Moscow, 125167, Russia Lyudmila Lyudmila Department of Corporate Finance and Corporate Governance, Financial University under the Government of the Russian Federation, Moscow, 125167, Russia Svetlana Dreving Department of Corporate Finance and Corporate Governance, Financial University under the Government of the Russian Federation, Moscow, 125167, Russia Olga Borisova Department of Corporate Finance and Corporate Governance, Financial University under the Government of the Russian Federation, Moscow, 125167, Russia Tatiana Tazikhina The expanding growth of financial transactions has resulted in the development of fraud systems. These progressions have considerably improved overall productivity, improved corporate management, and reduced operational costs. With the expanded utilization of automated financial transaction, organization and businesses have progressed to digital platform, convert their financial operation. Still, such a change in addition revealed financial systems to new threats, mainly through fraudulent activity and cybercrime. The large datasets, incorporated with the limits of conventional fraud detection techniques, provide a chance to accept Artificial Intelligence (AI) methods. The fraud detection problem is addressed by using Explainable AI (XAI) to give specialists with explained AI predictions over different explanation models. This paper proposes a Financial Transaction Fraud Detection via Dimensionality Reduction with an Explainable Artificial Intelligence Approach (FTFD-DRXAIA) technique. The aim is to develop an effective and intelligent system for accurate fraud detection in financial transaction utilizing progressive deep learning (DL) methods. Initially, the min-max method is used for data pre-processing to convert raw data into an appropriate format. Furthermore, the recursive feature elimination (RFE) system is applied for feature selection. For financial fraud detection process, the Elman recurrent neural network (ERNN) has been utilized. Moreover, the wildebeest herd optimization (WHO) method fine-tunes the ERNN model's hyperparameters, resulting in improved classification performance. Finally, the XAI technique applies LIME and SHAP to interpret complex AI models, enabling auditors and analysts to detect suspicious transaction patterns with greater clarity and confidence. The experimental outcome of FTFD-DRXAIA system is examined under the financial fraud detection database. The comparison analysis of FTFD-DRXAIA algorithm demonstrated an optimum precision value of 98.96% over recent methods. 2026 2026 08 22 10.54216/IJNS.270202 https://www.americaspg.com/articleinfo/21/show/4000