Enhancing Modern Malware Detection By Integrating LSTM And GAN

DOI

10.2991/978-94-6463-858-5_259How to use a DOI?

Keywords

LSTM - Long Short-term Memory; GAN – Generative Adversarial Networks; LF-Loss function; Malware Detection; Graph convolutional networks; Cybersecurity; Network Traffic Analysis

Abstract

The rapid evolution of malware poses significant challenges for traditional detection methods, necessitating advanced approaches that combine deep learning with effective interpretability techniques. This work combines Long Short Term Memory (LSTM) networks and Generative Adversarial Networks (GANs) to detect malware, which boosts the conventional rule based and signature based methods. Utilizing deep learning models, the system is capable of identifying both known and unknown malware better, lowering false positives and enhancing classification accuracy. One major component of this work is the use of visualization methods, such as confusion matrices and bar chart plots, to improve the interpretability of malware detection outcomes. The confusion matrix offers insights into classification performance by identifying correct and incorrect predictions, whereas the bar chart visually displays the number of correctly and incorrectly classified samples, allowing it to be easily assessed by cybersecurity analysts. These graphical tools improve decision making and enhance the transparency of AI-based malware detection systems. The project entails data preprocessing, feature extraction, and augmentation with GANs to provide diverse malware representation. The LSTM model learns sequential dependencies in malware execution patterns, and GANs generate synthetic malware data to enhance generalization. The real-time integration of visualization allows security analysts to monitor malware behavior dynamically, identify anomalies, and develop better insights into threat patterns. Performance is evaluated using accuracy, precision, recall, and F1-score metrics, with heavy visualization-based validation. The integration of deep learning models and visual analytics makes this framework very effective for malware classification, forensic analysis, and proactive cybersecurity. The proposed system not only maximizes malware detection accuracy but also closes the gap between AI driven models and human interpretability, rendering it a useful tool in today’s cybersecurity operations.

Copyright

© 2025 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - B. Sunayana
AU  - N. GuneshN
AU  - D.  Suryanarayana
AU  - M. Anitha
AU  - B. Komala Sai
PY  - 2025
DA  - 2025/11/04
TI  - Enhancing Modern Malware Detection By Integrating LSTM And GAN
BT  - Proceedings of International Conference on Computer Science and Communication Engineering (ICCSCE 2025)
PB  - Atlantis Press
SP  - 3097
EP  - 3111
SN  - 2352-538X
UR  - https://doi.org/10.2991/978-94-6463-858-5_259
DO  - 10.2991/978-94-6463-858-5_259
ID  - Sunayana2025
ER  -