Security and Privacy Analysis in Federated Active Learning for Supply Chain Management
Pages 1-13
https://doi.org/10.22034/lss.2026.564160.1060
Sattar Gheiratmand, MohammadAli Afshar Kazemi, Soheila Jokar, Erfaneh Noroozi
Abstract In the modern digital economy, the protection of privacy and security in data sharing has become a major concern, particularly within supply chains that rely on extensive data exchange between stakeholders. As supply chains evolve, the integration of advanced technologies like artificial intelligence (AI) and machine learning has revolutionized how companies predict demand, manage inventory, and optimize operations. This paper investigates the use of federated learning in supply chain management to address privacy and efficiency concerns. Federated learning allows decentralized data processing across multiple nodes, ensuring data privacy while maintaining high model accuracy. By employing privacy-preserving techniques such as differential privacy and encryption, the proposed model safeguards sensitive information from adversarial attacks, including model inversion and backdoor threats. The study also demonstrates the model’s effectiveness in reducing communication overhead, making it suitable for distributed supply chain systems. Although the findings are promising, further research is needed to optimize privacy-accuracy trade-offs, especially when dealing with non-IID data.
















