.png){kind=icon}
.png){kind=icon}
Return
Advanced malware detection system: enhancing security through CNN based inception ResNetV2 with transfer learning
.png){kind=icon}
PRE
AI
S
S
C
DOI：10.1007/s13198-026-03270-4.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Social media and malicious software both have been gaining popularity with different intents over the past decade. Social media provides a way to interact with intellectuals, while malicious software poses a threat to regular users on these platforms. The attacker deploys this malware in such a way that it is difficult to detect and becomes a security threat to user. After the entire software lifecycle is complete, then only victim gets aware about an attack happened; therefore, early-stage detection of malicious software is necessary to safeguard the system and ensure the security. The proposed work aims to determine a new way to classify malware using integrated convolutional neural network (CNN) with InceptionResNetV2 via transfer learning. The executable files in the dataset are converted into images, and the malware is then classified using the CNN+ InceptionResNetV2 model, which offers a unique ensemble method for merging the.byte and.asm data. At next stage, classification is done using transfer learning. The proposed algorithm is compared with other state-of-the-art algorithms like K-nearest neighbor, support vector machine, CNN+ Xception and others. Experimental analysis demonstrate that InceptionResNetV2 with transfer learning achieved the highest accuracy as compared to other methods, with 99.55% and 99.72% in both the datasets.
Keywords:
Malware analysis
Convolutional neural network
Transfer learning
Classification
Journal
I
IF:
1.4
Papers:
291
Citations:
3.0K