A High-Performance Deep Learning-Based Intrusion Detection System for Accurate Identification of Sophisticated Cyberattacks in Modern Network Environments
Jul 2026· African Journal Of Applied Research· Vol 12, pp. 26-40· 0 citations· 29 references
TL;DR
The experimental results show that the hybrid deep learning architecture outperforms individual models in network intrusion detection, and the hybrid model combining CNN, LSTM, and GRU achieved near-perfect accuracy, with extremely low false-positive and false-negative rates.
Abstract
Purpose: The purpose of this study is to design a reliable and high-performance intrusion detection system (IDS) that can effectively identify various sophisticated cyber-attacks in network environments using a hybrid deep learning approach.
Design/Methodology/Approach: A descriptive and experimental research methodology was used based on the UNSW NB15 benchmark dataset, which includes real and synthetic network data and various types of attacks. Data preprocessing includes handling missing values, encoding features, normalisation, and selecting features for dimensionality reduction. The performance of the models is evaluated using metrics such as accuracy, precision, recall, F1-score, and ROC-AUC.
Research Limitation: The study is limited to experiments conducted on the UNSW-NB15 dataset, and real-time deployment constraints such as computational overhead and resource limitations were not extensively evaluated.
Findings: The experimental results show that the hybrid deep learning architecture outperforms individual models in network intrusion detection. The hybrid model combining CNN, LSTM, and GRU achieved near-perfect accuracy in network intrusion detection, with extremely low false-positive and false-negative rates. The performance of recurrent models, such as LSTM, is superior in identifying network intrusion patterns, and the hybrid model performs best.
Practical Implication: The proposed hybrid IDS framework can be effectively used in real-world network infrastructures to improve proactive threat detection, minimise false negatives, and enhance cybersecurity defences against evolving attack patterns.
Social Implication: Improved intrusion detection systems help create a safer digital ecosystem by ensuring data safety, service availability, and trust in services delivered through networks, which is important for modern society.
Originality / Value: The current research provides a comprehensive hybrid deep learning framework for intrusion detection that leverages both feedforward and recurrent neural networks. It emphasises the power of fusion models in developing accurate and reliable intrusion detection systems, making it a valuable contribution for researchers and practitioners in this field.
An Enhanced Multi-Model Ensemble Network Intrusion Detection System (EME-NIDS), a deep meta-learning system that combines five different heterogeneous learning paradigms, including Convolutional Neural Networks, Dense Neural Networks, Transformers, XGBoost, and Random Forests is introduced.
The rapid growth of digital communication technologies, cloud computing, and Internet of Things (IoT) devices has
increased both the frequency and sophistication of cyber-attacks, making effective intrusion detection an essential component of
modern cybersecurity systems. Traditional signature-based intrusion detection systems (IDS) are effective against known attacks
but fail to detect previously unseen or evolving threats. This study investigates the application of deep learning models for binary
network intrusion detection using the NSL-KDD benchmark dataset. Three standalone architectures, Convolutional Neural
Networks (CNN), Long Short-Term Memory (LSTM) networks, and Deep Neural Networks (DNN), are implemented and
evaluated, alongside a CNN-LSTM Hybrid model that integrates spatial and sequential learning, and a DNN-LSTM Ensemble
model that combines independently trained DNN and LSTM predictions through weighted averaging. Following data cleaning,
categorical encoding, normalization, and Random Forest-based feature selection (41 features reduced to 20), all models were
trained and evaluated under identical conditions using Accuracy, Precision, Recall, F1-Score, ROC-AUC, training time, and
inference time. The standalone DNN model achieved the best overall performance, with 80.98% accuracy, 97.08% precision,
68.66% recall, 80.43% F1-score, and 96.11% ROC-AUC, while also requiring the shortest training time (39.69 s). The CNNLSTM Hybrid model attained the highest precision (97.23%) but did not outperform the standalone architectures overall, and
the DNN-LSTM Ensemble produced balanced but not superior results. These findings indicate that carefully designed
standalone architectures can match or exceed the performance of more complex hybrid and ensemble models for binary
intrusion detection, while incurring substantially lower computational cost. The study contributes a controlled, commonframework comparison of five deep learning architectures and provides practical guidance for selecting computationally
efficient models for anomaly-based intrusion detection.
Ketki Naik, Sanjeev Ghosh· International Journal for Re...· 0 citations
An intelligent hybrid deep learning framework based on a combination of deep neural networks (DNNs) and random forests (RF) to ensure the security of 5G IIoT networks for use in critical areas such as smart factories, cyber-physical systems, power grids, and industrial automation.
Rohan Rajoriya, Shweta Chouksey· International journal of com...· 0 citations
Cyber attacks are growing in number and complexity. Modern networks faces various real cyber threats such as API, DDPS, ICMP,UDP, TCP, botnet, Bit LINK kind of attacks. Intrusion detection system depends on machine learning for detect these attacks, but they faces various challenges in present scenario like un wanted data , poisoned data , stale data and privacy risk. Machine unlearning MU provides reliable and trust full solution by allowing various kind of latest model to remove harmful, unwanted, outdated data. This paper presents comprehensive survey of recent studies on machine UN learning applied to intrusion detection system IDS. We analyzed various approaches for unlearning time optimize, model accuracy, attacks types, and computational efficiency. the study highlight bets practices , performance trends, research gaps, time optimization , model performance accuracy , providing a roadmap for future development of high-accuracy, adaptive IDS frameworks. This paper provides researchers and practitioners with: (1) a structured, critical appraisal of the MU-IDS landscape; (2) quantitative benchmarks for cross-method comparison; (3) identification of unresolved challenges and adversarial threat models; and (4) concrete future research directions toward practical, privacy-compliant, and adversarially robust intrusion detection systems.
Sarmistha Podder, Saptarshi Paul· International journal of com...· 0 citations
This study examines a one-dimensional Convolutional Neural Network and a hybrid model, investigating how both architectures can detect network attacks in binary and multiclass classification settings, and provides actionable insights for practitioners choosing between deep learning and classical approaches under real-world NIDS deployment constraints.
LSTM had good detection for frequent attacks and slow-changing patterns, which shows its capacity in learning long-lasting dependencies, which shows its capacity in learning long-lasting dependencies.
Jawad Hussain Awan, Misbah Safdar, Muhammad Ayaz Shirazi et al.· Italian National Conference...· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
