Researchers Create AI For Complex Kolam Patterns

Researchers Shyamala and Seshadri Sivakumar have developed an AI algorithm that generates complex one-stroke 'sikku' kolam patterns, the work published recently in Nature's Heritage Science. The approach uses a customised upper-lower triangular recurrent neural network and a 'zigzag training algorithm' to mimic muscle memory, producing patterns up to 501 points versus typical human limits of 19–31 dots. Authors suggest mathematical foundations could enable applications in secure wireless communications, image encryption, and protein structure representation while preserving cultural heritage.