Optimizing Convolutional Neural Networks with Fish Swarm Algorithms for Mitigating Financial Risks in the Digital Economy

Abstract

Due to the fast growth and development of the digital economy, the method of exchanging and transacting money has changed drastically. However, this change also poses a previously unseen threat to the economy, particularly during periods of unexpected financial disasters. Complex emergency risk avoidance tactics are necessary in these kinds of scenarios. This project aims to construct a robust framework that can forecast and mitigate the impact of sudden financial crises on the digital economy using an intelligent fish swarm-optimized ensemble convolution neural network [IFSO-ECNN]. The proposed Deep Learning (DL) approach constructs prediction models by integrating relevant socioeconomic factors, market indicators, and a wealth of historical financial data. The financial data was prepared by applying min-max normalization. The features were extracted from the preprocessed data using kernel principal component analysis (K-PCA). This study contributes to our knowledge of how IFSO-ECNN can help the digital economy better prepare for and recover from unexpected financial catastrophes. Using both historical and real-time data, the suggested framework shows that it is possible to make smart judgements and proactively reduce financial risks. Experimental results validate the validity of the IFSO-ECNN model at 97.14% accuracy, 95.23% sensitivity, 96.25% specificity, and a 95.51% F1-score, surpassing the previous baseline traditions. The new hybrid model is more efficient than the previous models and has a new impact on AI-powered financial risk forecasting. The evaluation provides policymakers, financial institutions, and enterprises with useful insights for developing more resilient strategies in the face of unforeseen economic shocks by integrating modern IFSO-ECNN methodologies with domain experience.