Applications, Challenges, Recommendations, and Future Directions of Additive Manufacturing with Transfer Learning

Authors

• Qing Liu, Mississippi State University
• Ayantha Senanayaka, Tennessee Tech University
• Nayeon Lee, Mississippi State University
• Sungkwang Mun, Mississippi State University
• Amin Amirlatifi, Mississippi State University
• Joe Jabour, U. S. Army Engineer Research and Development Center
• Thomas Arnold, U. S. Army Engineer Research and Development Center
• Maria Seale, U. S. Army Engineer Research and Development Center

ORCID

Amerlatifi: https://orcid.org/0000-0002-2225-2087

MSU Affiliation

James Worth Bagley College of Engineering; Department of Computer Science and Engineering; Center for Advanced Vehicular Systems; Dave C. Swalm School of Chemical Engineering

Creation Date

2026-07-30

Abstract

This review provides a comprehensive discussion of how transfer learning (TL) techniques can be integrated to enhance process monitoring in additive manufacturing (AM), with the aim of transforming the field. AM has emerged as a powerful approach to create complex geometries and facilitate rapid prototyping, addressing critical drawbacks of traditional manufacturing such as material wastage, extended production timelines, high tooling costs, and intricate post-processing. Although data-driven monitoring during the AM process has improved AM’s reliability and productivity, they continue to face significant challenges due to the inherently dynamic nature of the process, limited data availability, and increased post-processing costs. TL, a subset of machine learning that transfers knowledge from data-rich source domains to data-scarce targets, offers a promising solution. This paper begins by providing an overview of machine learning development in AM. It highlights the challenges faced within the AM domain, the limitations of AM modeling, and the importance of high-level monitoring of AM processes. It then outlines strategies for applying TL to tackle these issues and offers comprehensive insights into integrating TL frameworks in the context of AM. The review identifies three key AM trends in the TL application: anomaly detection, process monitoring, and geometric deviation modeling. Moreover, it examines data acquisition methods, dominant pre-trained architectures, and evaluation approaches within the scope of integrating TL in AM. Finally, we identify seven main challenges associated with TL implementation in AM and offer practical guidelines to address these challenges, thereby proposing viable paths for future AM advancement.

Publication Date

6-22-2026

Publication Title

Advanced Engineering Informatics

Publisher

Elsevier

Creative Commons License

This work is licensed under a Creative Commons Attribution 4.0 International License.

Rights

© 2026 The Authors

Recommended Citation

Liu, Q., Senanayaka, A., Lee, N., Mun, S., Amirlatifi, A., Jabour, J., Arnold, T., & Seale, M. (2026). Applications, challenges, recommendations, and future directions of additive manufacturing with transfer learning. Advanced Engineering Informatics, 76, 104962. https://doi.org/10.1016/j.aei.2026.104962