A malware detection method with function parameters encoding and function dependency modeling.

Area of Science:

• Computer Science
• Cybersecurity
• Machine Learning

Background:

• Malware poses a significant threat to network security in the digital age.
• Existing machine learning and deep learning methods for malware detection often overlook function parameter and dependency analysis.

Purpose of the Study:

• To develop an advanced malware detection technique that incorporates function parameter and dependency analysis.
• To enhance the accuracy and robustness of malware detection systems.

Main Methods:

• A parameter encoder was designed to transform function parameters into feature vectors.
• Clustering methods were employed for discretizing parameter features to improve API encoding.
• A deep neural network was utilized to capture functional dependencies and generate semantic representations of function sequences.

Main Results:

• The proposed method achieved 98.62% accuracy and a 98.40% F1-score on a large-scale dataset.
• Ablation experiments confirmed the critical importance of function parameters and dependencies in detection.
• The new technique demonstrated superior performance compared to existing methods.

Conclusions:

• The integration of function parameter and dependency analysis offers a more effective approach to malware detection.
• This research provides a robust method for improving network security against sophisticated malware.
• The findings highlight the potential of deep learning for advanced cybersecurity applications.