Skip to content
Open access

QuadFusion-IDS: Enhancing Intrusion Detection with Unified Hybrid Deep and Machine Learning Techniques in Cybersecurity Systems

Sep 2026 · Electronics · 0 citations · 47 references

TL;DR

QuadFusion-IDS provides a strong and interpretable binary intrusion detection approach for the UNSW-NB15 benchmark, while further validation under live traffic, adversarial conditions, streaming workloads, and operational cybersecurity environments remains necessary.

Abstract

As network traffic becomes increasingly complex and cyberattack behaviors continue to evolve, conventional intrusion detection approaches face challenges in accurately identifying malicious and legitimate network activities. This study proposes QuadFusion-IDS, a hybrid binary intrusion detection framework that integrates FT-Transformer with Autoencoder (FT-Transformer + AE), multi-head attention deep neural network (MHA-DNN), dilated temporal convolutional network (Dilated TCN), and XGBoost to capture complementary feature representations, temporal dependencies, and nonlinear traffic patterns. The framework employs a controlled preprocessing pipeline involving categorical encoding, variance-based filtering, correlation analysis, Random Forest-based feature selection, normalization, stratified train-validation-test partitioning, and ADASYN-based class balancing, while maintaining a locked test set to minimize information leakage. Using the UNSW-NB15 dataset, QuadFusion-IDS achieved an accuracy of 0.9545, a precision of 0.9566, a recall of 0.9775, an F1-score of 0.9669, an MCC of 0.8946, and an AUC-ROC of 0.9922 for binary classification of normal and attack traffic. Ablation experiments demonstrated the importance of ensemble components and optimized weighting, with XGBoost removal producing the largest reduction in F1-score and AUC-ROC. Multi-seed evaluation further showed small performance variations, supporting the stability of the proposed framework. Feature selection analysis indicated that the selected 20-feature configuration substantially reduces input dimensionality while retaining competitive detection performance compared with the original 42-feature configuration. SHAP and LIME analyses provide feature-level and local decision explanations for network traffic predictions. Overall, QuadFusion-IDS provides a strong and interpretable binary intrusion detection approach for the UNSW-NB15 benchmark, while further validation under live traffic, adversarial conditions, streaming workloads, and operational cybersecurity environments remains necessary.

Read PDF

Similar papers

Open access Aug 2026

HADS-Net: A Hybrid Attention-Based Deep Security Network for Network Intrusion Detection

The principal contribution of this work is architectural and diagnostic rather than a performance improvement: it documents that combining feature-wise attention with out-of-fold stacked generalization does not, in this setting, outperform a plain multi-layer perceptron, while incurring the highest memory footprint of...

Mahima Khanna, V. Murthy, Siva Ramavarapu et al. · 0 citations
Open access 2026

A LIGHTWEIGHT HYBRID DEEP LEARNING FRAMEWORK FOR MULTI-CLASS NETWORK INTRUSION DETECTION

The growing variety of cyberattacks has increased the need for intrusion detection systems that can detect multiple categories of attack with high reliability. We propose a lightweight hybrid framework LHCBA-Net, which integrates one dimensional convolutional neural networks, bidirectional long short term memory and te...

S. Mohan, Gowtham Reddy Kunduru, Shaik Jaffar Hussain et al. · 0 citations
Open access 2026

HYBRIDNET-IDS: AN INNOVATIVE CNN, BILSTM, AND TRANSFORMER FRAMEWORK UTILISING ATTENTION FOR IMPROVED NETWORK INTRUSION DETECTION

The rapid spread of network-connected technologies and the growing complexity of attacks have shown that traditional signature-based intrusion detection systems are not good enough, especially when it comes to finding zero-day attacks and advanced persistent threats. This paper introduces HybridNet-IDS, an augmented de...

Subhodip Koley, Sumit Das, Debasish Saha Roy et al. · 0 citations
Open access Aug 2026

An Enhanced Hybrid Deep Learning Model for Anomaly-Based Intrusion Detection in the CICIDS2017 Web Attack Traffic

An enhanced hybrid deep learning architecture that combines one-dimensional convolutional layers, bidirectional long short-term memory units, and a multi-head self-attention mechanism for detecting web attacks in network-flow data is proposed.

Israa Shihab Ahmed, Wasan Alaa Hussain, Z. H. Rasool · 0 citations
Open access 2026

A Hybrid Autoencoder and Transformer-Based Framework for Intrusion Detection in IoT Networks

: Intrusion detection in Internet-of-Things (IoT) and enterprise networks must combine high predictive performance with reliable minority-attack recognition, leakage-safe evaluation, and deployable computational cost. This study presents an Optimized Gated Autoencoder-Transformer that preserves standardized raw flow de...

A. Javed, A. Bakhshi, M. M. Alam et al. · 0 citations
Conference Aug 2026

Deepsecure-IDS: A Deep Convolutional Learning Framework for High-Performance Network Intrusion Detection

Recent advancements in network technology have led to the emergence of numerous sophisticated intrusion types that differ significantly from existing threats. Traditional firewalls cannot immediately detect these intrusions with their own set of rules and policies. Therefore, an Intrusion Detection System (IDS) is esse...

R. Raja, S. Vairachilai · 0 citations

We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.