Answers are grounded in this paper and cite their sources.
Demonstrates enhanced fraud detection in UPI transactions, indicating improved accuracy and interpretability.
Financial security is seriously at risk from Unified Payments Interface fraud since it is difficult for conventional fraud detection techniques to keep up with new fraud patterns. Hence, this research proposed a novel dilated convolutional-based Trans network for the UPI fraud detection approach, which is used to detect various types of transaction fraud. Initially, the input raw data are collected from UPI Transactions 2024, pre-processed using the min-max normalization approach, used fix range to improve convergence speed in detecting UPI fraud. After pre-processing, the data features are extracted using a dual-channel and spatial attention integrated convolutional class-wise autoencoder, effectively extracts a set of relevant features. The classification method uses a dilated convolutional max-pooled locality attention-based Trans network to address the long-range dependencies of significant features. Then, parameter optimization using the Randomly Aligned Dung Beetle Optimization to explore and exploit the process for fine-tuning hyperparameters. This algorithm is used to adjust the parameters, and the optimization prevents premature integration. Experimental findings of this study demonstrated the effectiveness of the proposed system, which attained 98.28% accuracy and an F1-score value of 98.26%. In this research, outcomes indication that the proposed system has enhanced interpretability, detects UPI fraud patterns, and improves model accuracy.
No takes yet. Share an insight, caveat, or question.
Pimprale et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: