Multi-Objective Optimization for Credit Card Fraud Classification: Achieving a Balance between Accuracy and Carbon Footprint within a Green AI Framework
Jul 2026· 2026 11th International Conference on Applying New Technology in Green Buildings (ATiGB)· pp. 673-678· 0 citations· 11 references
Abstract
As financial fraud becomes increasingly sophisticated, the demand for complex machine learning models has surged, inadvertently leading to a significant increase in computational energy consumption. This study addresses the critical trade-off between predictive accuracy and environmental sustainability within a Green AI framework. We propose a multi-objective optimization approach to evaluate Logistic Regression, Random Forest, and XGBoost on a highly imbalanced credit card fraud dataset. Experimental results reveal a non-linear “Carbon Cost of Complexity,” where the transition from linear to tree-based architectures yields diminishing returns; a 3.4% improvement in detection accuracy requires a 163% increase in training carbon emissions. The Random Forest model $(\mathbf{n}=\mathbf{2 0 0}, \mathbf{d}=\mathbf{1 5})$ emerged as the Pareto Optimal solution, achieving a superior F1-Score of 0.7512 and an AUPRC of 0.8031. Although XGBoost proved to be 62% more energy-efficient during the training phase, Random Forest demonstrated a distinct advantage in inference latency, achieving a throughput of 664,576 Transactions Per Second (TPS). We conclude that while Random Forest incurs a higher carbon footprint $\left(\mathbf{7. 1 5} \times \mathbf{1 0}^{-\mathbf{6}} \mathbf{~ k g C O} \mathbf{2 e q}\right.$ per cycle), this expenditure is justified by its robustness in preventing financial loss and its capability for real-time processing in high-stakes environments.
Abstract The rise in online transactions has made credit card fraud a significant global concern, necessitating detection strategies that are both highly accurate and practically viable. While existing literature extensively explores machine learning techniques to address class imbalance, most studies optimize for traditional statistical metrics, overlooking the asymmetric financial costs and strict operational constraints of real-world fraud detection. This study bridges this gap by proposing a comprehensive, cost-sensitive ensemble framework evaluated on a real-world European cardholder dataset. We move beyond the traditional F 1 score by adopting the cost-sensitive F β metric to reflect real financial impact. Through exhaustive benchmarking, we show that while eXtreme Gradient Boosting (XGBoost) combined with Borderline SMOTE achieves the highest single-model performance, our proposed soft-voting ensemble integrating Logistic Regression and Random Forest with SMOTE delivers the best overall performance (F β = 0.8287). To ensure practical viability, we introduce a Top-K operational constraint evaluation reflecting limited human investigation bandwidths. Additionally, an ablation study demonstrates that there is no universal remedy for class imbalance; optimal interventions are highly model-dependent. Finally, by validating our framework on a feature-transparent simulated dataset, model interpretability analysis reveals the ensemble’s capacity to capture the critical importance of environmental risk factors, shifting the focus beyond solely customer-centric anomalies.
Xin-Yue Fan, T. Boonen· Asia-Pacific Journal of Risk...· 0 citations
This work provides a mathematically grounded benchmarking framework for integrating Explainable Artificial Intelligence (XAI) into fraud detection pipelines, aligning high-accuracy analytics with the transparency requirements expected in regulated financial environments.
Henrique Barros, F. Antunes, Maryam Abbasi· International Conference on...· 0 citations
Credit card fraud detection presents unique challenges for machine learning due to extreme class imbalance, evolving fraud patterns, and asymmetric misclassification costs. While numerous algorithms have been proposed for this domain, their evaluation typically relies on simplistic metrics that fail to capture the multifaceted requirements of operational fraud detection systems. This study introduces a sophisticated multi‐criteria ranking framework that extends beyond traditional performance measures to incorporate temporal stability, computational efficiency, and robustness considerations. We develop an anisometric penalty structure that quantifies deficiencies across multiple dimensions with differential weighting, and we apply this framework to evaluate several leading machine learning approaches with various sampling strategies, XGBoost, Random Forest, and logistic regression—using a large dataset of European credit card transactions. Our findings reveal that an active learning approach with combined uncertainty–diversity sampling achieves superior performance across multiple evaluation criteria, outperforming both XGBoost and traditional classification algorithms. Statistical significance testing and sensitivity analysis confirm the robustness of these results across different weight configurations and operational scenarios. This study advances both the theoretical understanding of machine learning evaluation in fraud detection and provides practical guidance for financial institutions seeking to implement or enhance their fraud detection systems.
M. Zahid Yüzügüldü, H. Altún, Ali Coşkun et al.· International Transactions i...· 0 citations
Credit card fraud has become a central issue for the financial security of banks and cardholders. However, fraudulent transactions account for only a very small proportion of all transaction cases. If it is missed, it may directly cause huge property damage. As a result, this study focuses on how to identify fraudulent transactions as much as possible, while controlling false positives and human audit costs. Specifically, logistic regression, gradient boosting, XGBoost, and random forest are compared, while class weighting, SMOTE, and Random Undersampling are evaluated for handling class imbalance. It is more critical to translate the model results into a risk warning system and a loss simulation system. The results show that random forest with class weighting has the best overall performance, with 90.6% F1-Score, 94.1% precision, and 87.3% recall. Meanwhile, under the cost assumptions, this model reduces simulation costs by 85.4% relative to the no-model baseline. These findings suggest that in data environments with severe class imbalances, a machine learning model can be designed as an effective triage tool. Its actual value is not only to predict fraud, but also to establish a low-risk release, medium-risk verification, and high-risk manual review of the decision-making process for institutions and truly reduce losses.
Zi-Yue Meng· Advances in Economics, Manag...· 0 citations
Accurate detection of financial fraud remains a critical challenge due to information asymmetry, high-dimensional data complexity, and evolving fraudulent behaviors. This study develops an artificial-intelligence-based financial fraud identification framework for listed companies by integrating multiple machine-learning algorithms. A financial indicator database covering profitability, solvency, cash flow, and governance characteristics is first constructed. Principal Component Analysis is employed to reduce dimensionality and eliminate multicollinearity, followed by the training of Random Forest, Support Vector Machine, and Neural Network models. A fusion-learning strategy with cross-validation optimization is further introduced to enhance model robustness and classification performance. Experimental results demonstrate that the fusion model achieves an accuracy of 94.7%, outperforming individual RF (83.2%), SVM (85.6%), and NN (89.1%) models while maintaining a false-positive rate below 5%. The proposed framework provides effective support for intelligent risk warning and financial supervision and offers methodological references for pattern recognition, data fusion, and intelligent decision-making systems.
The study introduces a Stacked Logistic Regression ensemble to combine the predictive capacity of optimized Random Forest and XGBoost base classifiers and reveals that the proposed stacked model performance surpasses both individual base models.
Uduh Israel Akakoh, G. N. Edegbe· FUDMA Journal of Sciences· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
