Poster: Intrusion Detection System Based on Federated Transfer Learning
Y. Hu, Guoming Zhang · 2023
Anomaly detection methods based on machine learning (ML) have been widely used in intrusion detection systems (IDS). However, the majority of existing studies rely on a centralized training approach, which requires the collection of data from each user, posing a privacy risk. In this paper, we present a federated learning-based intrusion detection method that uses each user’s data to train the model, thereby ensuring privacy. In addition, we integrate transfer learning by using publicly available data during the training process to alleviate computational resource constraints on individual nodes. Experimental results validate the effectiveness of our proposed method, demonstrating a remarkable accuracy rate of 99.57%. These results highlight the potential of our approach to improving intrusion detection performance while mitigating privacy concerns and addressing resource constraints.