AI-Driven Defense Mechanisms for Protecting Industry 5.0 from Android Malware Threats
ADITYA ADITYA, Chandra Sekhar Dash · 2024
This study seeks to establish the level of accuracy of CNN, RNN, and GNN in the classification of Android malware threats in industrial 5. 0 applications. To notice the strengths and weaknesses of these neural network architectures the following evaluation metrics are used: precision, recall, F1-score, and accuracy. From the obtained results, it is seen that though CNNs provide fairly good accuracy and recall and log loss in most of the cases, RNNs succeed in sequential data analysis, and GNNs are significantly higher in precision and accuracy in case with graph data. Such conclusions reveal the applicability of GNNs to the tasks that call for fine-grained analysis of relations and high detection accuracy of malware. Thus, the research helps to advance knowledge on the strategic implementation of AI-based countermeasures to secure Industry 5. 0 from evolving cyber threats with a focus on GNNs’ applied relational learning.