The reliability of historical and cultural primary sources, and hence the scientific value, depends on the degree of study. In the case of a huge number and variety of petroglyphs and related cultural artifacts, it is necessary to conduct a complex, multifaceted classification which will facilitate the extraction of functions, semantic and spatio-temporal aspects of the petroglyphs. Modern Machine Learning (ML) technologies are also applicable in Petroglyphology. Their contribution is promising in the problems of image recognition and classification.ML is a subset of artificial intelligence that automatically enables a machine (neural network) or system to learn and improve from experience. Instead of explicit programming, it uses algorithms to make decisions based on the analysis of large amounts of data. When exposed to more data, the model’s performance improves, ultimately enhancing its accuracy and reliability.Using a Supervised Learning approach, one of the main types of ML, the model is trained on a set of labeled images provided by the researcher. Supervised Learning involves pairing each petroglyph with its content, allowing the trained model to accurately predict the content of new, previously unseen images. Unsupervised Learning can also be applied when the model is trained on data that have not been labeled by the researcher.This will contribute to solving a number of fundamental problems of the Rock-art of Armenia because it is distinguished to a certain extent by its antiquity, multi-genre content and, especially, applied significance. Machine Learning methods developed for petroglyphs can be applied to other spheres of cultural studies improving image recognition, analysis, and data classification.

KeywordsArmenia archaeology database artificial intelligence classification Machine Learning petroglyph Rock-art