Attention-Enhanced CNN–ResNet with XGBoost Ensem-ble and DAMCE Loss for Intrusion Detection

Bousba, Billel and Ihmeida, Mohamed (2025) Attention-Enhanced CNN–ResNet with XGBoost Ensem-ble and DAMCE Loss for Intrusion Detection. In: International Conference of Reliable Information and Communication Technology, 18th-20th Nov 2025, Morocco. (In Press)

Text IRICT_2025_Camera_Ready_Paper_40.pdf - Accepted Version Restricted to Repository staff only Download (336kB) | Request a copy

Abstract

Intrusion detection is a critical component of modern cybersecu-rity yet building systems that are both accurate and generalisable remains a sig-nificate challenge due to diverse attack types, imbalanced datasets, and evolving threat patterns. This study proposes a hybrid intrusion detection framework evaluated on the UNSW-NB15, CICIDS2017, and NSL-KDD datasets. The framework combines a one-dimensional convolutional neural network en-hanced with attention and residual blocks for feature generalisation, with an XGBoost classifier tailored for tabular decision boundaries. The preprocessing pipeline integrates categorical encoding, mutual information feature selection, synthetic oversampling, and feature standardisation. To address the limitations of conventional objective functions, we introduce the Difficulty-Aware MSE+CE (DAMCE) loss, which adaptively balances cross-entropy and mean squared error based on sample difficulty. Experimental results demonstrate that the proposed ensemble consistently outperforms standalone models, while DAMCE improves stability and class-wise calibration compared to standard loss functions. These findings underline the value of integrating deep learning with boosting and highlight the importance of adaptive loss design in develop-ing robust and generalisable intrusion detection systems. Code link on GitHub: IRICT/CNN-Xgboost.ipynb at main · BillBs-13/IRICT.

Item Type:	Conference or Workshop Item (Paper)
Dates:	Date Event 25 November 2025 Accepted
Uncontrolled Keywords:	Intrusion Detection System (IDS), UNSW-NB15, CICIDS2017, NSL-KDD, Convolutional Neural Network (CNN), Attention Mechanism, Residual Learning, XGBoost,; Cybersecurity.
Subjects:	CAH11 - computing > CAH11-01 - computing > CAH11-01-01 - computer science
Divisions:	Architecture, Built Environment, Computing and Engineering > Computer Science
Depositing User:	Gemma Tonks
Date Deposited:	02 Feb 2026 13:39
Last Modified:	02 Feb 2026 13:39
URI:	https://www.open-access.bcu.ac.uk/id/eprint/16830

Actions (login required)

View Item