A Hierarchical Learning Approach to Detect Aging Faces based on Local Binary Pattern Evaluation and Optimization Logics

Aging face detection is a critical area in facial recognition research, with applications spanning security, forensic analysis, and age-based authentication. This study presents a Hierarchical Learning Approach to Detect Aging Faces based on Local Binary Pattern (LBP) Evaluation and Optimization Logics. The proposed model leverages LBP-based texture extraction, Genetic Algorithm (GA) for feature selection, and a Hybrid CNN-LSTM classifier with an attention mechanism to improve prediction accuracy. The system was trained and evaluated on publicly available face aging datasets, demonstrating superior performance compared to traditional classifiers. Experimental results indicate that the CNN-LSTM hybrid model with attention mechanism achieved an accuracy of 92.4%, significantly outperforming conventional LBP-SVM (81.2%), LBP-XGBoost (85.3%), and standalone CNN models (87.9%). The integration of Bayesian Optimization and Focal Loss further enhanced classification robustness and reduced bias in age group prediction. Feature selection using GA reduced the feature space by 70.7% while improving accuracy to 91.2%, demonstrating its effectiveness. This work provides a computationally efficient and highly accurate approach for aging face detection, making it suitable for real-world applications such as age verification systems, criminal identification, and social security authentication.