An Deep Learning-Based Malicious Code Detection System Combining Graph Convolutional Networks and Bidirectional LSTM

The rapid evolution of cyber threats demands more effective detection systems that can adapt to complex and dynamic network environments. This paper introduces a novel deep learning-based approach that combines Graph Convolutional Networks (GCN) and Bidirectional Long Short-Term Memory (BiLSTM) to enhance the detection of malicious code. By leveraging both the spatial relationships within network data and the temporal dynamics of traffic flows, the proposed model achieves high accuracy in detecting diverse types of malicious activities. Experiments conducted on the CICIDS2018 dataset demonstrate that the hybrid model outperforms traditional deep learning models that rely solely on either graph-based or sequence-based methodologies. The results indicate significant improvements in accuracy, precision, recall, and F1-score, suggesting that the integration of GCN and BiLSTM provides a robust framework for future cybersecurity solutions.