Revolutionizing Temperament Assessment: An Investigation into Inception-V3 Architecture Applied to Dermatoglyphics

Drawing from the intersection of artificial intelligence and behavioral psychology, this research investigates the efficacy of a deep neural network, specifically utilizing the Inception-v3 architecture, in accurately classifying fingerprint patterns and their corresponding temperaments. Employing an experimental design, the study utilized a synthetic self-generated fingerprint dataset using Anguli, achieving a noteworthy 91.78% average accuracy in fingerprint pattern classification. Notably, the model exhibited consistent performance despite variations in noise or scratch levels added to the dataset. To establish the link between temperaments and fingerprint patterns, 120 participants underwent the Eysenck Personality Inventory (EPI) assessment under the supervision of a registered psychometrician. Then, all fingerprints of every participant were captured for temperament association. The subsequent analysis indicated a moderate overall accuracy (M=51%) in temperament classification. Nevertheless, the model consistently achieved a high accuracy score (M=80%) in identifying melancholic temperament, particularly associated with the whorl-dominant fingerprint pattern. These results provide comprehensive insights into the capabilities of Inception-v3 in fingerprint pattern classification and its potential for temperament assessment, thereby contributing to advancements in the understanding of the intersection between AI and behavioral psychology.