Table 2 Summary of related work on obfuscated malware detection.
From: Transfer learning with XAI for robust malware and IoT network security
Refs. | Focus | Dataset | Findings & results | Limitations |
|---|---|---|---|---|
Graph Neural Networks (GNN) for obfuscated malware detection | Custom dataset based on malware graph structures | Achieved 94.3% accuracy in detecting obfuscated malware | Requires large labeled datasets; computationally expensive | |
Impact of obfuscation on malware detection techniques | Multiple malware datasets, including public repositories | Showed significant drop in detection accuracy for traditional methods | Lack of a proposed mitigation strategy; limited real-world testing | |
Smart memory forensics for Windows malware detection | Memory dumps from Windows devices | Demonstrated 92% accuracy using memory analysis techniques | Focuses only on Windows devices; lacks comparison with other OS | |
Machine learning for obfuscated malware detection in memory dumps | Public and synthetic memory dump datasets | Improved detection rates compared to traditional heuristics | May suffer from adversarial attacks; requires frequent retraining | |
Real-world obfuscated malware detection through memory analysis | Memory snapshots of real-world malware samples | Achieved over 90% detection accuracy in various scenarios | Performance may vary with unseen malware samples; potential overfitting | |
Explainable AI for obfuscated malware detection | Lightweight memory-based dataset | XMal model achieved competitive results with lower resource consumption | Limited interpretability for complex obfuscation techniques | |
Privacy-focused malware detection via memory dumping analysis | Large-scale memory dump dataset | Effective classification with minimal false positives | High computational cost; privacy concerns with memory analysis | |
Malware detection using machine learning models | Various malware repositories | Compared multiple ML models, with deep learning achieving the highest accuracy | Feature selection requires refinement; high false positive rate | |
Deep autoencoders for malware detection using memory analysis | Temporal evaluation-based dataset | Stacked ensemble model achieved over 95% accuracy | Model performance depends on proper hyperparameter tuning | |
Memory feature engineering for obfuscated malware detection | Experimental dataset from controlled memory environments | Demonstrated effective feature engineering for malware detection | Requires extensive feature extraction; high dependency on dataset quality |
