Table 1.
Comparative analysis of android malware detection techniques.
Fig 1.
Overall Block diagram of the Proposed model.
Fig 2.
Flow of Feature Vector Generator.
Fig 3.
Overall Flow diagram of the ProposedDBN-GRU model.
Finally, we create that the hybrid DBN-GRU model architecture simultaneously extracts unsupervised features for Android applications with sequential dependency model using GRU to enhance the analysis of Android applications in a comprehensive manner. We demonstrate this approach can identify sophisticated malware patterns and significantly improve detection accuracy in dynamic Android environments.
Table 2.
Dataset composition.
Table 3.
Performance comparison of proposed and baseline approaches based on metrics.
Table 4.
Preprocessing accuracy levels.
Table 5.
Feature extraction time levels (in seconds).
Table 6.
Feature selection accuracy levels.
Table 7.
Feature dimensionality reduction accuracy levels.
Table 8.
DBN-GRU feature processing accuracy levels.
Table 9.
Malware detection time levels (in seconds).
Fig 4.
Comparative Performance Metrics Analysis of DBN-GRU and Traditional models.
Fig 5.
AUC Comparison Curve for Different models.
Fig 6.
Preprocessing Accuracy Levels.
Fig 7.
Feature Extraction Time Levels.
Fig 8.
Feature Selection Accuracy Levels.
Fig 9.
Feature Dimensionality Reduction Accuracy Levels.
Fig 10.
DBN-GRU Feature Processing Accuracy Levels.
Fig 11.
Malware Detection Time Levels.
Table 10.
Malicious signal detection accuracy levels.
Fig 12.
Malicious Signal Detection Accuracy Levels.