Federated Learning for Linux Malware Detection: An Experimental Study

The key to the success of Artificial Intelligent models is data. However, a major challenge is that data security and privacy issues must be considered. Federated Learning (FL) is considered an approach to overcoming challenges of data sensibility. In this study, we perform experiments with the Federated Learning model on detecting malware in the Linux operating system. We compare Federated Learning with distributed data and traditional Fully Connected Models with centralized data. Although the Federated Learning model has not been optimized for the configured parameters, the Linux malware detection results are very positive compared to the traditional model. In particular, the Federated Learning model has very high data security.