- Title
- Malware detection in edge devices with fuzzy oversampling and dynamic class weighting
- Creator
- Khoda, Mahbub; Kamruzzaman, Joarder; Gondal, Iqbal; Imam, Tasadduq; Rahman, Ashfaqur
- Date
- 2021
- Type
- Text; Journal article
- Identifier
- http://researchonline.federation.edu.au/vital/access/HandleResolver/1959.17/179522
- Identifier
- vital:15623
- Identifier
-
https://doi.org/10.1016/j.asoc.2021.107783
- Identifier
- ISBN:1568-4946 (ISSN)
- Abstract
- In Internet-of-things (IoT) domain, edge devices are used increasingly for data accumulation, preprocessing, and analytics. Intelligent integration of edge devices with Artificial Intelligence (AI) facilitates real-time analysis and decision making. However, these devices simultaneously provide additional attack opportunities for malware developers, potentially leading to information and financial loss. Machine learning approaches can detect such attacks but their performance degrades when benign samples substantially outnumber malware samples in training data. Existing approaches for such imbalanced data assume samples represented as continuous features and thus can generate invalid samples when malware applications are represented by binary features. We propose a novel malware oversampling technique that addresses this issue. Further, we propose two approaches for malware detection. Our first approach uses fuzzy set theory, while the second approach dynamically assigns higher priority to malware samples using a novel loss function. Combining our oversampling technique with these approaches, the proposed approach attains over 9% improvement over competing methods in terms of F1_score. Our approaches can, therefore, result in enhanced privacy and security in edge computing services. © 2021 Elsevier B.V.
- Publisher
- Elsevier Ltd
- Relation
- Applied Soft Computing Vol. 112, no. (2021), p.
- Rights
- All metadata describing materials held in, or linked to, the repository is freely available under a CC0 licence
- Rights
- Copyright ©2021 Elsevier B.V.
- Subject
- 0102 Applied Mathematics; 0801 Artificial Intelligence and Image Processing; 0806 Information Systems; Dynamic class weighting; Fuzzy oversampling; Imbalanced data; Malware detection in edge devices
- Reviewed
- Funder
- This research is supported by Australian Government Research Training Program (RTP) Stipend and RTP Fee-Offset Scholarship through Federation University Australia. It is also supported by Data61-CSIRO, Australia.
- Hits: 823
- Visitors: 635
- Downloads: 0
| Thumbnail | File | Description | Size | Format |
|---|