Artificial intelligence-based antivirus in order to detect malware preventively

The proposed paper investigates commercial antiviruses. About 17% of the antiviruses did not recognize the existence of the malicious samples analyzed. In order to overcome the limitations of commercial antiviruses, this project creates an antivirus able to identify the modus operandi of a malware a...

Full description

Saved in:
Bibliographic Details
Published in:Progress in artificial intelligence 2021-03, Vol.10 (1), p.1-22
Main Authors: de Lima, Sidney M. L., Silva, Heverton K. de L., Luz, João H. da S., Lima, Hercília J. do N., Silva, Samuel L. de P., de Andrade, Anna B. A., da Silva, Alisson M.
Format: Article
Language:English
Subjects:
Anti-virus software
Artificial Intelligence
Artificial neural networks
Computational Intelligence
Computer Imaging
Computer Science
Control
Data Mining and Knowledge Discovery
Empirical analysis
Feature extraction
Machine learning
Malware
Mechatronics
Natural Language Processing (NLP)
Neural networks
Pattern Recognition and Graphics
Regular Paper
Response time
Robotics
Statistical methods
Vision
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:The proposed paper investigates commercial antiviruses. About 17% of the antiviruses did not recognize the existence of the malicious samples analyzed. In order to overcome the limitations of commercial antiviruses, this project creates an antivirus able to identify the modus operandi of a malware application before it is even executed by the user. In the proposed methodology, the features extracted from the executables are the input attributes of artificial neural networks. The classification of neural networks aims to group executables of 32-bit architectures into two classes: benign and malware. In total, 6272 executables are used in order to validate the proposed methodology. The proposed antivirus achieves an average performance of 98.32% in the distinction between benign and malware executables, accompanied by an average response time of only 0.07 s. Our antivirus is statistically superior and more effective when compared to the best state-of-the-art antivirus. The limitations of commercial antiviruses can be catering for artificial intelligence techniques based on machine learning. Instead of empirical and heuristic models, the proposed work identifies, in a statistical way, behaviors previously classified as suspects in real time.
ISSN:2192-6352
2192-6360
DOI:10.1007/s13748-020-00220-4