A Novel Hybrid Ensemble Framework with RealTime Evolutionary Optimization for Enhancing Biometric Authentication

This paper presents a hybrid ensemble learning framework for facial biometric authentication that combines Snapshot Ensembles, Dynamic Ensemble Selection (DES), and Adaptive Evolutionary Ensemble Learning (AEEL) to address the limitations of traditional ensemble methods (Bagging, Boosting, Stacking) in handling high-dimensional biometric data, noise, and real-time responsiveness. Six machine learning classifiers (SVM, Random Forest, XGBoost, KNN, Decision Tree, and Neural Network) were evaluated across six ensemble strategies on a facial biometric dataset. Results show that AEEL achieved the highest average accuracy among all methods, with its best-performing model, XGBoost, further optimized using evolutionary algorithms (Genetic Algorithms, Particle Swarm Optimization, and Ant Colony Optimization) to boost performance without the computational overhead of tuning every model. The framework was validated across multiple datasets, confirming its robustness and adaptability for real-time, secure identity verification systems in finance, healthcare, and surveillance applications.