Pseudo-Image Representation for CNN-Based Intrusion Detection
Abstract
Convolutional Neural Networks (CNNs) have demonstrated strong potential for Intrusion Detection Systems (IDSs); however, their performance is highly dependent on input data representation. This paper presents a lightweight pseudo image transformation framework that converts network traffic into optimized two-dimensional layouts for CNN processing, focusing on data representation rather than architectural modifications. Evaluated on NSL-KDD and CIC-IDS2017 datasets, square and near-square formats are shown to enhance spatial feature extraction. In particular, an 11×11 base representation, upscaled to 64×64 using bicubic interpolation, achieves a precision of 99.2%, accelerates convergence, and remains robust under limited training data conditions. By avoiding complex hybrid architectures, the proposed approach reduces computational overhead while maintaining competitive performance, making it suitable for real-time, large-scale deployment. These results suggest that strategic data representation, rather than increased model complexity, can significantly improve CNN-based IDS performance.
Keywords
Convolutional Neural Networks (CNNs), Data Representation, Intrusion Detection Systems (IDSs), Pseudo Images, Cybersecurity, NSL-KDD, CIC-IDS2017
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.
H. Chaibi and A. Marref, "Pseudo-Image Representation for CNN-Based Intrusion Detection," in Journal of Communications Software and Systems, vol. 22, no. 4, pp. 619-630, August 2026, doi: 10.24138/jcomss-2026-0085
@article{chaibi2026pseudoimage,
author = {Chaibi, Hassene and Marref, Amine},
title = {{Pseudo-Image Representation for CNN-Based Intrusion Detection}},
journal = {Journal of Communications Software and Systems},
month = aug,
year = {2026},
volume = {22},
number = {4},
pages = {619--630},
doi = {10.24138/jcomss-2026-0085},
url = {https://doi.org/10.24138/jcomss-2026-0085}
}