Android Malware Detection Using Machine Learning with Feature Selection Based on the Genetic Algorithm
Since the discovery that machine learning can be used to effectively detect Android malware, many studies on machine learning-based malware detection techniques have been conducted. Several methods based on feature selection, particularly genetic algorithms, have been proposed to increase the perfor...
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2021-11-01
|
| Series: | Mathematics |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2227-7390/9/21/2813 |
| _version_ | 1827678123660935168 |
|---|---|
| author | Jaehyeong Lee Hyuk Jang Sungmin Ha Yourim Yoon |
| author_facet | Jaehyeong Lee Hyuk Jang Sungmin Ha Yourim Yoon |
| author_sort | Jaehyeong Lee |
| collection | DOAJ |
| description | Since the discovery that machine learning can be used to effectively detect Android malware, many studies on machine learning-based malware detection techniques have been conducted. Several methods based on feature selection, particularly genetic algorithms, have been proposed to increase the performance and reduce costs. However, because they have yet to be compared with other methods and their many features have not been sufficiently verified, such methods have certain limitations. This study investigates whether genetic algorithm-based feature selection helps Android malware detection. We applied nine machine learning algorithms with genetic algorithm-based feature selection for 1104 static features through 5000 benign applications and 2500 malwares included in the Andro-AutoPsy dataset. Comparative experimental results show that the genetic algorithm performed better than the information gain-based method, which is generally used as a feature selection method. Moreover, machine learning using the proposed genetic algorithm-based feature selection has an absolute advantage in terms of time compared to machine learning without feature selection. The results indicate that incorporating genetic algorithms into Android malware detection is a valuable approach. Furthermore, to improve malware detection performance, it is useful to apply genetic algorithm-based feature selection to machine learning. |
| first_indexed | 2024-03-10T05:57:12Z |
| format | Article |
| id | doaj.art-cc88efdfb8064a069f3b63b14def0b80 |
| institution | Directory Open Access Journal |
| issn | 2227-7390 |
| language | English |
| last_indexed | 2024-03-10T05:57:12Z |
| publishDate | 2021-11-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Mathematics |
| spelling | doaj.art-cc88efdfb8064a069f3b63b14def0b802023-11-22T21:19:17ZengMDPI AGMathematics2227-73902021-11-01921281310.3390/math9212813Android Malware Detection Using Machine Learning with Feature Selection Based on the Genetic AlgorithmJaehyeong Lee0Hyuk Jang1Sungmin Ha2Yourim Yoon3Department of Computer Engineering, Gachon University, 1342 Seongnamdaero, Sujeong-gu, Seongnam-si 13120, Gyeonggi-do, KoreaDepartment of Computer Engineering, Gachon University, 1342 Seongnamdaero, Sujeong-gu, Seongnam-si 13120, Gyeonggi-do, KoreaDepartment of Business Administration, Gachon University, 1342 Seongnamdaero, Sujeong-gu, Seongnam-si 13120, Gyeonggi-do, KoreaDepartment of Computer Engineering, Gachon University, 1342 Seongnamdaero, Sujeong-gu, Seongnam-si 13120, Gyeonggi-do, KoreaSince the discovery that machine learning can be used to effectively detect Android malware, many studies on machine learning-based malware detection techniques have been conducted. Several methods based on feature selection, particularly genetic algorithms, have been proposed to increase the performance and reduce costs. However, because they have yet to be compared with other methods and their many features have not been sufficiently verified, such methods have certain limitations. This study investigates whether genetic algorithm-based feature selection helps Android malware detection. We applied nine machine learning algorithms with genetic algorithm-based feature selection for 1104 static features through 5000 benign applications and 2500 malwares included in the Andro-AutoPsy dataset. Comparative experimental results show that the genetic algorithm performed better than the information gain-based method, which is generally used as a feature selection method. Moreover, machine learning using the proposed genetic algorithm-based feature selection has an absolute advantage in terms of time compared to machine learning without feature selection. The results indicate that incorporating genetic algorithms into Android malware detection is a valuable approach. Furthermore, to improve malware detection performance, it is useful to apply genetic algorithm-based feature selection to machine learning.https://www.mdpi.com/2227-7390/9/21/2813android malware detectionmachine learninggenetic algorithmfeature selectionstatic analysis |
| spellingShingle | Jaehyeong Lee Hyuk Jang Sungmin Ha Yourim Yoon Android Malware Detection Using Machine Learning with Feature Selection Based on the Genetic Algorithm Mathematics android malware detection machine learning genetic algorithm feature selection static analysis |
| title | Android Malware Detection Using Machine Learning with Feature Selection Based on the Genetic Algorithm |
| title_full | Android Malware Detection Using Machine Learning with Feature Selection Based on the Genetic Algorithm |
| title_fullStr | Android Malware Detection Using Machine Learning with Feature Selection Based on the Genetic Algorithm |
| title_full_unstemmed | Android Malware Detection Using Machine Learning with Feature Selection Based on the Genetic Algorithm |
| title_short | Android Malware Detection Using Machine Learning with Feature Selection Based on the Genetic Algorithm |
| title_sort | android malware detection using machine learning with feature selection based on the genetic algorithm |
| topic | android malware detection machine learning genetic algorithm feature selection static analysis |
| url | https://www.mdpi.com/2227-7390/9/21/2813 |
| work_keys_str_mv | AT jaehyeonglee androidmalwaredetectionusingmachinelearningwithfeatureselectionbasedonthegeneticalgorithm AT hyukjang androidmalwaredetectionusingmachinelearningwithfeatureselectionbasedonthegeneticalgorithm AT sungminha androidmalwaredetectionusingmachinelearningwithfeatureselectionbasedonthegeneticalgorithm AT yourimyoon androidmalwaredetectionusingmachinelearningwithfeatureselectionbasedonthegeneticalgorithm |