Hybrid GA-CS Optimized Deep Learning Framework for Fraud Detection in Ethereum Blockchain Transactions

Abstract

The rapid growth of Ethereum smart contracts has attracted significant attention from both academia and industry, leading to the emergence of diverse commercial applications. However, the increasing prevalence of fraudulent activities such as phishing, bribery, and money laundering poses serious threats to the security and integrity of online transactions. To address these challenges, this study proposes a deep learning-based fraud detection framework enhanced with a novel metaheuristic optimization technique. Specifically, an Optimized Genetic Algorithm-Cuckoo Search (GA-CS) hybrid approach is integrated with a deep learning model to improve fraud classification accuracy. The proposed GA-CS algorithm leverages the global search capability of Cuckoo Search while employing Genetic Algorithm operations to overcome its inherent limitations and enhance convergence performance. Extensive experiments are conducted to evaluate the effectiveness of the proposed method against several widely used machine learning and deep learning classifiers, including Logistic Regression (LR), K-Nearest Neighbors (KNN), MultiLayer Perceptron (MLP), XGBoost, Light Gradient Boosting Machine (LGBM), Random Forest (RF), and Support Vector Classification (SVC), using a limited feature set. Experimental results demonstrate that the proposed GA-CS-optimized deep learning model outperforms most baseline methods and achieves superior detection accuracy. Although its performance is marginally higher than that of the Random Forest model, the proposed approach and the SVC model achieve the highest overall accuracy, confirming the robustness and effectiveness of the proposed framework for fraudulent transaction detection on the Ethereum platform.