Description of Invention
This study explores Federated Learning (FL) as a defense against cybersecurity threats. It introduces an efficient FL system, combining Convolutional Neural Networks (CNN) and Long Short-Term Memory (LSTM) models for improved intrusion detection while maintaining data privacy. Experiments with varying numbers of clients (5, 10, and 15) using Dynamic Weighted Aggregation Federated Learning (DWAFL) show high accuracy (92.2% with 5 clients, 94.2% with 10 clients, and 93.2% with 15 clients). FL with DWAFL holds promise for precise intrusion detection, preserving data confidentiality, and advancing collaborative learning systems for security applications.