Search engine for discovering works of Art, research articles, and books related to Art and Culture
ShareThis
Javascript must be enabled to continue!

AN EFFECTIVE MALWARE CLASSIFICATION METHOD BASED ON BYTE-TO-IMAGE TRANSFORMATION AND INTEGRATION OF THE VISION TRANSFORMER MODEL

View through CrossRef
Malware classification is a critical problem in cybersecurity, characterized by numerous challenges due to the complexity and diversity of malware variants. In this study, we propose a novel approach that transforms bytecode into image representations and employs the Vision Transformer (ViT) architecture for malware family classification. The proposed data preprocessing method preserves essential structural information of the malware while simplifying feature extraction. ViT leverages the self-attention mechanism to model complex and long- range dependencies, offering advantages over traditional CNN-based models. Experiments conducted on the Microsoft Malware Classification Challenge dataset demonstrate that the proposed model achieves high accuracy and F1-scores, particularly for malware families such as Kelihos_ver3 and Lollipop. Confusion matrix analysis reveals a strong discriminative capability across malware families, while also highlighting challenges in distinguishing families with structurally similar or heavily obfuscated patterns. The study also discusses current limitations, including computational cost and the lack of integration of dynamic behavioral data, and outlines future research directions to improve performance and real-world applicability. Overall, the results highlight the potential of Vision Transformer architectures in malware classification, suggesting a promising avenue for further research in cybersecurity. Keywords: Malware classification; vision transformer (ViT); bytecode image representation; deep learning; self-attention.
Title: AN EFFECTIVE MALWARE CLASSIFICATION METHOD BASED ON BYTE-TO-IMAGE TRANSFORMATION AND INTEGRATION OF THE VISION TRANSFORMER MODEL
Description:
Malware classification is a critical problem in cybersecurity, characterized by numerous challenges due to the complexity and diversity of malware variants.
In this study, we propose a novel approach that transforms bytecode into image representations and employs the Vision Transformer (ViT) architecture for malware family classification.
The proposed data preprocessing method preserves essential structural information of the malware while simplifying feature extraction.
ViT leverages the self-attention mechanism to model complex and long- range dependencies, offering advantages over traditional CNN-based models.
Experiments conducted on the Microsoft Malware Classification Challenge dataset demonstrate that the proposed model achieves high accuracy and F1-scores, particularly for malware families such as Kelihos_ver3 and Lollipop.
Confusion matrix analysis reveals a strong discriminative capability across malware families, while also highlighting challenges in distinguishing families with structurally similar or heavily obfuscated patterns.
The study also discusses current limitations, including computational cost and the lack of integration of dynamic behavioral data, and outlines future research directions to improve performance and real-world applicability.
Overall, the results highlight the potential of Vision Transformer architectures in malware classification, suggesting a promising avenue for further research in cybersecurity.
Keywords: Malware classification; vision transformer (ViT); bytecode image representation; deep learning; self-attention.

Related Results

Image-Based Malware Multiclass Classification Using Vision Transformer Architecture
Image-Based Malware Multiclass Classification Using Vision Transformer Architecture
Perkembangan malware yang semakin canggih telah menjadi ancaman serius bagi keamanan siber global, mengakibatkan kerugian finansial yang signifikan. Metode deteksi tradisional sepe...
Dynamic Features for Robust Malware Detection: A Systematic Review, Taxonomy, and Practical Analysis Framework
Dynamic Features for Robust Malware Detection: A Systematic Review, Taxonomy, and Practical Analysis Framework
The need to mitigate malware attacks cannot be overemphasized, as they pose serious threats to the critical information assets in cyberspace. Understanding and utilizing appropriat...
MCPDS: image-based malware classification method using PE metadata alone
MCPDS: image-based malware classification method using PE metadata alone
Abstract In response to the increasing threat posed by the exponential growth of malware in cybersecurity, researchers have developed a numbe...
Malware Behaviour Visualization
Malware Behaviour Visualization
The number of unique malware variants released each year is on the rise. Researchers may often need to use manual static and dynamic analysis to study new malware samples. Manual a...
Automatic Load Sharing of Transformer
Automatic Load Sharing of Transformer
Transformer plays a major role in the power system. It works 24 hours a day and provides power to the load. The transformer is excessive full, its windings are overheated which lea...
High frequency modeling of power transformers under transients
High frequency modeling of power transformers under transients
This thesis presents the results related to high frequency modeling of power transformers. First, a 25kVA distribution transformer under lightning surges is tested in the laborator...
Depth-aware salient object segmentation
Depth-aware salient object segmentation
Object segmentation is an important task which is widely employed in many computer vision applications such as object detection, tracking, recognition, and ret...
An optimal deep learning-based framework for the detection and classification of android malware
An optimal deep learning-based framework for the detection and classification of android malware
 The use of smartphones is increasing rapidly and the malicious intrusions associated with it have become a challenging task that needs to be resolved. A secure and effective techn...

Back to Top