Table 1.
Summary of existing techniques.
Table 2.
Comparison of proposed model with existing techniques.
Fig 1.
Proposed model of Malware visualization and classification.
Fig 2.
Dataset creating steps.
Fig 3.
28 classes sample images of private dataset.
Fig 4.
Graphical representation of the staking-ensemble model.
Table 3.
Private dataset description.
Table 4.
Cross-dataset generalization.
Table 5.
Ablation study model backbone and Fusion on private and Malimg dataset.
Table 6.
Class-wise performance on Private dataset.
Fig 5.
Family-wise confusion matrices for Adware, Trojan, Worm, and Win32 families.
Fig 6.
ROC curve comparison across Malware Families.
Fig 7.
Meta learning performance across Epochs.
Table 7.
Class Class-wise accuracy of baseline models using Quadratic SVM classifier.
Table 8.
Zero-day attack detection on Malware Families.
Table 9.
Zero-day detection with leave-one-family-out protocol (averaged over 5 seeds).
Table 10.
Performance of Private dataset on deep model and classifiers.
Table 11.
Performance metric of AlexNet, ResNet-18 and ResNet-50 on each node using Custom dataset.
Fig 8.
Four different confusion matrix results on AlexNet and ResNet-18.
Table 12.
Performance metric of AlexNet, ResNet-18 and ResNet-50 on each node using Malimg dataset.
Fig 9.
Node level comparison of different deep models.
Table 13.
Comparison of Malimg dataset and Private dataset.