Article: Zero-sample accounting standards migration framework empowered by meta-learning Journal: International Journal of Information and Communication Technology (IJICT) 2026 Vol.27 No.62 pp.27 - 45 Abstract: With the global economy's deep integration, traditional manual migration methods cannot meet efficiency and accuracy needs. This study explores meta learning techniques' application in zero sample accounting standard transfer and constructs an intelligent framework to adapt to new accounting standards in various fields. By analysing meta learning mechanisms, an innovative transfer framework is designed to use a small amount of source domain data for rapid adaptation and precise transfer of accounting standards in new fields. Experimental results show that the meta learning empowerment framework significantly improves transfer performance under zero sample conditions. Compared with traditional methods, the average accuracy is up by 12.3%. At different data scales, as the data volume increases from 100 to 1,000, the accuracy improves by 8.5%, 10.2%, and 13.1% respectively. In new accounting standard testing, the average accuracy reaches 85.6%, a 9.4% improvement over traditional methods. Inderscience Publishers - linking academia, business and industry through research

Title: Zero-sample accounting standards migration framework empowered by meta-learning

Authors: Xianyi Xiong; Shibing Zhang

Addresses: Hunan Vocational College of Commerce, Changsha, 410205, China ' Hunan Vocational College of Commerce, Changsha, 410205, China

Abstract: With the global economy's deep integration, traditional manual migration methods cannot meet efficiency and accuracy needs. This study explores meta learning techniques' application in zero sample accounting standard transfer and constructs an intelligent framework to adapt to new accounting standards in various fields. By analysing meta learning mechanisms, an innovative transfer framework is designed to use a small amount of source domain data for rapid adaptation and precise transfer of accounting standards in new fields. Experimental results show that the meta learning empowerment framework significantly improves transfer performance under zero sample conditions. Compared with traditional methods, the average accuracy is up by 12.3%. At different data scales, as the data volume increases from 100 to 1,000, the accuracy improves by 8.5%, 10.2%, and 13.1% respectively. In new accounting standard testing, the average accuracy reaches 85.6%, a 9.4% improvement over traditional methods.

Keywords: meta-learning; zero-sample learning; accounting standards migration; intelligent framework.

DOI: 10.1504/IJICT.2026.153994

International Journal of Information and Communication Technology, 2026 Vol.27 No.62, pp.27 - 45

Received: 08 Dec 2025
Accepted: 19 Jan 2026
Published online: 09 Jun 2026 *

Title: Zero-sample accounting standards migration framework empowered by meta-learning

Keep up-to-date