Evolving with Klinkenberg’s Idea (EKI) Algorithms for Automatic Identification of Sa Huynh Antique Glass Artifacts

Abstract

This research focuses on the challenge of comprehensively identifying Sa Huynh glass jewelry in Vietnamese archaeology based on new SEM gemological analysis. We propose the implementation of a system, called Recognition Automatic System for Sa Huynh Glasses (RAS-SHG), motivated by the unique conditions in archaeology, that aims to employ evolving continuous learning algorithms on our Sa Huynh Culture archaeological databases. This research develops new evolving continuous learning algorithms, so called Evolving with Klinkenberg's Idea (EKI) algorithms, based on a combination of different classic machine learning algorithms with the "sliding windows" approach and Klinkenberg's optimized window size. We compare these algorithms to select the most suitable model that aligns with the performance requirements of the Sa Huynh Culture archaeological dataset. This Recognition Automatic System for Sa Huynh Glasses (RAS-SHG) has been developed to accurately distinguish between Sa Huynh and non-Sa Huynh glass ornaments and is now utilized at the Vietnam Institute of Archaeology and the UNESCO Centre for researching and preserving Vietnamese antiquities.