The application of computational intelligence in genetic improvement offers valuable support to breeders in the decision-making process. This study aims to assess the potential of computational intelligence for exploring genetic diversity associated with seed yield performance in safflower genotypes. Conducted in 2022 at the University of Maragheh, Iran, the experiment evaluated 95 safflower accessions arranged in an alpha lattice design with two replications. Morphological, agronomic, and yield-related traits were assessed, with emphasis on seed yield performance. Using Kohonen’s self-organizing maps (SOM), the genotypes were categorized into 11 clusters based on a network configuration of four columns and three rows. Predictor Importance analysis of the SOM network indicated that among 18 measured traits, the weight of the main capitulum, number of capitula per main branch, and number of lateral branches per plant were the most influential on seed yield. Other key traits included number of seeds per main capitulum, stem diameter, diameter of lateral capitulum, and number of main branches per plant. The results demonstrate that computational intelligence, particularly SOM, is a powerful and efficient tool for uncovering genetic diversity patterns linked to yield performance in safflower. These traits show strong potential for use in breeding programs aimed at cultivar development in semi-arid regions.
