About this Abstract |
| Meeting |
2025 Annual International Solid Freeform Fabrication Symposium (SFF Symp 2025)
|
| Symposium
|
2025 Annual International Solid Freeform Fabrication Symposium (SFF Symp 2025)
|
| Presentation Title |
Federated Learning for Defect Detection in Additive Manufacturing: Mitigating Label-Flipping Attacks in Distributed Factories |
| Author(s) |
Md Sazol Ahmmed, Sriram Praneeth Isanaka, Atiqur Rahman, Frank Liou |
| On-Site Speaker (Planned) |
Atiqur Rahman |
| Abstract Scope |
In modern manufacturing, Distributed Digital Factories (DDFs) are a revolutionary concept that combines several geographically separate factories into a cooperative system. Though this distributed nature of DDFs has many advantages, it also creates certain difficulties, such as sharing data for defect identification in Additive Manufacturing (AM). Traditional methods in AM are data-intensive, requiring large, high-quality datasets, which are difficult to obtain across diverse, non-IID (Independent and Identically Distributed) data from various factories. This research has presented Federated Learning (FL) models as a solution by enabling collaborative model training without sharing sensitive data. However, FL in DDFs faces vulnerability to label-flipping attacks. To counter this, this paper also proposes a secure, privacy-preserving FL framework tailored for AM in DDFs. The proposed framework improves AM machine model accuracy, security, and adaptability by allowing factories to collaborate on defect detection. |
| Proceedings Inclusion? |
Planned: Post-meeting proceedings |