Review of Transfer Learning Techniques for Cybersecurity in Digital Supply Chain Systems: Models, Applications, and Challenges

Abstract

The increasing complexity of digital supply chain systems introduces dynamic cybersecurity challenges that traditional machine learning approaches often fail to address due to their reliance on domain-specific labeled data. Transfer Learning (TL) has emerged as a promising paradigm to overcome these limitations by enabling knowledge reuse across domains, thereby enhancing cyber threat detection and adaptability. Guided by the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) framework, this review systematically examines TL-based approaches applied to cybersecurity within digital supply chain environments. The study identifies key TL paradigms domain adaptation, inductive learning, transductive learning, and deep transfer learning and analyzes their roles in securing critical components of the digital ecosystem. The findings categorize research efforts into four major focus areas: network security, information security, web application security, and Internet of Things (IoT) security. Furthermore, the review highlights a significant surge in publications from 2020 onward, reflecting the growing scholarly and practical interest in TL-driven cybersecurity solutions. It concludes by identifying existing research gaps and proposing future directions for developing adaptive and intelligent cross-domain cybersecurity frameworks aimed at strengthening the resilience of digital supply chain systems.