Generative AI based Interpretation of Human Ultrametrics and Analysis

This paper presents a biometric identification system based on tongue print recognition with the help of machine learning techniques. The system is named Generative-AI-based Interpretation of Human Ultrametrics and Analysis (GIHUA). Traditional biometric systems like fingerprints, facial recognition and iris scans suffer the challenges of computational complexity, forgery risks and age-related changes. Tongue print, on the contrary has the advantages that it remains unchanged throughout the life of a person, unaffected by trauma or diseases and is very difficult to forge, due to difficulty in its acquisition without the awareness of the individual. Tongue-print based systems mostly rely on image-based dataset whereas this work focuses on a numerical dataset which contributes to computational efficiency. GIHUA uses ultrametric distances to extract the features and this makes it ideal for applying directly into any machine learning model, regardless of being linear or non-linear. To overcome the scarcity of labeled tongue-print datasets, GIHUA creates synthetic labels with the help of K-means clustering and generative AI-based data augmentation technique. Classification is performed using random forest algorithm. Results show an accuracy of 99.10%, promising tongue print as an additional or alternative approach to biometric identification.