Explainable Artificial Intelligence-Based IoT Device Malware Detection Mechanism Using Image Visualization and Fine-Tuned CNN-Based Transfer Learning Model

Table 5

Comparison of the proposed model with previously published works.

Published works (year)	Detection model	Feature type	Data source	Accuracy (%)
Yen et al. [18]	Convolutional neural network	Code retrieved from APKs with image-based characteristics	—	92
Ullah et al. [28]	Deep convolutional neural network	Code retrieved from APKs with image-based characteristics	R2-D2 IoT device dataset	97.46
de Oliveira et al. [20]	Combined CNN + DNN + TN neural network model	Characteristics extracted from static and dynamic analyses	Omnidroid	90.90
Hamad et al. [21]	K-nearest neighbor	Code retrieved from APKs with image-based characteristics	R2-D2 IoT device dataset	97.29
Hamad et al. [29]	Deep convolutional neural network	Code retrieved from APKs with image-based characteristics	R2-D2 IoT device dataset and malimg	98
Mercaldo et al. [22]	Deep neural network	Code retrieved from binaries with image-based characteristics	Google play store and AMD	91.8
Yadav et al. [30]	Pretrained efficient net convolutional neural network	Code retrieved from binaries with image-based characteristics	R2-D2 IoT device dataset	95.7
Proposed approach	Pretrained Inception-v3	Code retrieved from APKs with image-based characteristics	R2-D2 IoT device dataset	98.5