What question did this study set out to answer?

This research aims to develop an efficient malware detection model suitable for resource-constrained environments.

June 12, 2026International Journal of Data and Network Science1 citationsOpen Access

Lightweight malware detection model in a resource-limited environment using single-head attention with LSTM/GRU

MGMariam Al GhamriAmman Arab University AEAfaf EdinatUniversity of Jordan MSMohammad Shehab

Key Points

This research aims to develop an efficient malware detection model suitable for resource-constrained environments.
Introduced a network combining single-head attention mechanism with LSTM or GRU.
Utilized the CIC-MalMem-2022 dataset.
Focused on learning temporal features from memory data.
Achieved an accuracy of 92%.
Reduced training time by 30%.
Enhanced model efficiency for real-life applications.

Abstract

The rising malware threats in the past years are due to the proliferation of smart devices and resource-constrained systems, such as the Internet of Things (IoT) and mobile devices. This phenomenon creates great difficulties for conventional defense mechanisms, which frequently lag behind because of changes in the malware landscape. In this work, a lightweight and powerful network combining a single-head attention mechanism with LSTM or GRU for this task, based on the CIC-MalMem-2022 dataset, is introduced. The focus is on learning temporal features retrieved from memory data, considering model efficiency for resource-constrained devices. Results indicate that the generated model can retain an accuracy of 92%, and it can save training time by 30%, which is highly beneficial for time-critical tasks and efficient resource usage in real-life applications. This model contributes to the enhancement of cybersecurity by offering good practice against the growing range of new threats in today's technology-advanced environments.

Ask AI

Helpful

Bookmark

View Full Paper