Security Impact of Federated and Transfer Learning on Network Management Systems with Fuzzy DEMATEL Approach

Abstract

: Everyday using of the big data, machine learning algorithms, and related studies, ensuring data privacy and security have become a critical necessity. These features make them more vulnerable to cyber-attacks. The security of the stored data is also critical, and evaluating the processing of information in the autonomous network management of these systems. The criteria considers the account in the processing and security of data entering every field from the widespread industry examined. It is necessary to increase their awareness of negative and attack problems while these systems are working. Applications such as traditional machine learning and the use of cloud computing also involve risks regarding data security and personal data leakage. Cooperative learning pays due attention to the confidentiality of sensitive information by keeping the original training data hidden. By collecting, combining, and integrating heterogeneous data with collaborative learning together with a federated learning structure, data produced and stored. This study discusses the effect of federated and transfer learning on autonomous network management analyzes the security status parameters. The fuzzy DEMATEL method was preferred in exploring the parameters affecting the system state according to the degree of importance. Situational scenarios evaluated by considering the structure in which the features of cyber-physical systems examined together with federated learning. Data security factors discussed with the fuzzy DEMATEL