An automatic updating perceptron-based system for malware detection
Published in 2013 15th International Symposium on Symbolic and Numeric Algorithms for Scientific Computing, 2013
Recommended citation: Barat, Marius and Prelipcean, Dumitru Bogdan and Gavrilut, Dragos Teodor, "An automatic updating perceptron-based system for malware detection." 2013 15th International Symposium on Symbolic and Numeric Algorithms for Scientific Computing, pages 303-307, IEEE, 2013. https://doi.org/10.1109/SYNASC.2013.48
Abstract
This work introduces an automatic updating perceptron-based system specifically designed for malware detection. The system features adaptive learning capabilities that allow it to continuously improve its detection accuracy by automatically incorporating new threat patterns.
Key Contributions
- Automatic Updates: Self-updating mechanism that adapts to new malware patterns
- Perceptron Architecture: Efficient neural network approach for binary classification
- Adaptive Learning: Continuous improvement without manual intervention
- Scalable Design: System architecture that handles large-scale malware datasets
Technical Innovation
The system implements:
- Dynamic weight adjustment based on new malware samples
- Automated feature extraction from executable files
- Real-time learning algorithms for threat pattern recognition
- Efficient memory management for large-scale deployment
Industry Impact
This research contributed to the development of adaptive malware detection systems at Bitdefender, enabling automatic updates to detection models without requiring manual retraining.