.png){kind=icon}
.png){kind=icon}
Return
Efficient hybrid feature selection using intrinsic and metaheuristic optimization algorithm and classification of malware using ensemble learning algorithm
.png){kind=icon}
PRE
AI
DOI：10.1007/s10586-025-05743-w.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Malware, a type of software designed to destroy, gain unauthorized access, or cause damage to computer systems, puts sensitive information and data at risk in cyberspace. Despite numerous studies, there still remains a major gap in building reliable systems capable of optimized, accurate, and scalable malware detection. This paper focuses on the key area of malware feature identification and selection using various feature selection algorithms, which include intrinsic methods and metaheuristic optimization to improve the accuracy of malware classification. An experimental and comparative analysis of existing systems has been conducted to explore a novel hybrid model for effective feature selection. Based on the evaluation results, the best-performing algorithms were identified. Initially, least absolute shrinkage and selection operator Cross-Validation (Lasso CV) was employed for primary feature filtering, followed by the application of the metaheuristic SHADE (Success-History based Adaptive Differential Evolution) algorithm for advanced feature selection. This hybrid approach achieved an accuracy of 97.5% and an AUC of 99%, effectively addressing local optima challenges. The proposed model demonstrates a robust and efficient framework for accurate malware detection and classification.
Keywords:
Meta-heuristic
Ensemble learning
Malware
Genetic algorithm
Local optima
Deep learning.
Journal
C
IF:
0
Papers:
691
Citations:
1
View more organizations