Designing an expert system for ranking facial attractiveness based on deep learning and fuzzy inference system

Facial attractiveness plays a crucial role in various social contexts, influencing mate selection, social interactions, hiring decisions, fashion modeling, and marketing strategies. This paper presents the design and development of an expert system for ranking facial attractiveness, leveraging fuzzy inference systems (FIS). Initially, we conducted a comprehensive review of existing research on facial attractiveness, revealing a gap in the application of expert systems for attractiveness assessment. To address this gap, we integrated deep learning techniques, specifically a Convolutional Neural Network (CNN), to identify attractive faces. Subsequently, we employed fuzzy inference system rules to construct an expert model for the precise ranking of facial attractiveness. Our study demonstrates the exceptional effectiveness of combining FIS with CNN in ranking facial attractiveness. When evaluated on a limited dataset, our method outperforms other existing approaches. The proposed model achieves impressive results with mean and standard deviation values of sensitivity (99.46±0.03%), specificity (99.35±0.01%), precision (98.99±0.02%), and accuracy (99.7±0.01%) in classifying attractive and unattractive faces. Furthermore, our approach yields a mean prediction accuracy of 99.53+0.06% for ranking facial attractiveness. In this work, we have developed an expert ranking system for facial attractiveness that categorizes faces based on structural features. Our approach not only provides a practical tool for attractiveness assessment but also contributes to a deeper understanding of how FIS and CNN can effectively rank facial attractiveness.