Home / Current Issue / Paper 1710529
Explainable Deep Neural Networks for Real-Time Malware Classification in Enterprise Systems
Subject area: Science,Engineering and Technology · Area of research: Cyber Security
DOI: https://doi.org/10.64388/IREV9I3-1710529-439
Abstract
The rise and growing complexity of malware present a serious and ongoing threat to enterprise systems. Traditional methods that rely on signatures for detection just aren't cutting it anymore when it comes to dealing with polymorphic and zero-day threats. Enter deep neural networks (DNNs), which have proven to be a robust solution, boasting high accuracy and the capability to identify new malware variants by learning intricate patterns from extensive datasets. However, their "black-box" nature?meaning we can't easily understand how they make decisions?can be a barrier to their use in critical enterprise security situations. This paper introduces a thorough framework for real-time malware classification using explainable deep neural networks (XDNNs) tailored for enterprise environments. We suggest an architecture that combines a high-performance deep learning model with post-hoc explainability techniques like SHapley Additive exPlanations (SHAP) and Local Interpretable Model-agnostic Explanations (LIME). Our method analyzes both static and dynamic malware features to achieve impressive detection accuracy while also giving security analysts valuable insights into the model's decision-making process. We assess the trade-offs between model performance, computational demands, and the clarity of explanations, showing that XDNNs can strike a crucial balance between effectiveness and interpretability, ultimately fostering trust and enhancing the operational efficiency of a security operations center (SOC).
Keywords
Explainable AI (XAI), Deep Learning, Deep Neural Networks (DNNs), Malware Classification, Real-time Malware Detection, Enterprise Security, Cybersecurity.
How to cite this paper
@article{1710529,
author = {Dr. Deepak Tomar, Dr. Kismat Chhillar, Prof. Alok Verma},
title = {Explainable Deep Neural Networks for Real-Time Malware Classification in Enterprise Systems},
journal = {Iconic Research And Engineering Journals},
year = {2025},
volume = {9},
number = {3},
pages = {496-502},
issn = {2456-8880},
url = {https://www.irejournals.com/formatedpaper/1710529.pdf},
abstract = {The rise and growing complexity of malware present a serious and ongoing threat to enterprise systems. Traditional methods that rely on signatures for detection just aren't cutting it anymore when it comes to dealing with polymorphic and zero-day threats. Enter deep neural networks (DNNs), which have proven to be a robust solution, boasting high accuracy and the capability to identify new malware variants by learning intricate patterns from extensive datasets. However, their "black-box" nature?meaning we can't easily understand how they make decisions?can be a barrier to their use in critical enterprise security situations. This paper introduces a thorough framework for real-time malware classification using explainable deep neural networks (XDNNs) tailored for enterprise environments. We suggest an architecture that combines a high-performance deep learning model with post-hoc explainability techniques like SHapley Additive exPlanations (SHAP) and Local Interpretable Model-agnostic Explanations (LIME). Our method analyzes both static and dynamic malware features to achieve impressive detection accuracy while also giving security analysts valuable insights into the model's decision-making process. We assess the trade-offs between model performance, computational demands, and the clarity of explanations, showing that XDNNs can strike a crucial balance between effectiveness and interpretability, ultimately fostering trust and enhancing the operational efficiency of a security operations center (SOC).},
keywords = {Explainable AI (XAI), Deep Learning, Deep Neural Networks (DNNs), Malware Classification, Real-time Malware Detection, Enterprise Security, Cybersecurity.},
month = {September},
doi = {https://doi.org/10.64388/IREV9I3-1710529-439}
}