A Novel Two-Stage Deep Learning Model for Network Intrusion Detection: LSTM-AE

IEEE Access, Journal Year: 2023, Volume and Issue: 11, P. 37131 - 37148

Published: Jan. 1, 2023

Machine learning and deep techniques are widely used to assess intrusion detection systems (IDS) capable of rapidly automatically recognizing classifying cyber-attacks on networks hosts. However, when destructive attacks becoming more extensive, challenges develop, needing a comprehensive response. Numerous datasets publicly accessible for further analysis by the cybersecurity research community. no previous has examined performance proposed model variety in detail. Due dynamic nature attack its changing techniques, must be updated benchmarked regularly. The neural network (DNN) convolutional (CNN) this article as types models developing flexible effective IDS detecting comparing them with cyber-attacks. constant development behavior fast growth need evaluation many produced over time through static methods. This kind enables identification most efficient algorithm identifying future We novel two-stage technique hybridizing Long-Short Term Memory (LSTM) Auto-Encoders (AE) attacks. CICIDS2017 CSE-CICDIS2018 determine optimum parameters LSTM-AE. experimental results show that hybrid works well is applicable modern scenarios.

Language: Английский

---

Enhancing IoT network security through deep learning-powered Intrusion Detection System

Internet of Things, Journal Year: 2023, Volume and Issue: 24, P. 100936 - 100936

Published: Sept. 13, 2023

The rapid growth of the Internet Things (IoT) has brought about a global concern for security interconnected devices and networks. This necessitates use efficient Intrusion Detection System (IDS) to mitigate cyber threats. Deep learning (DL) techniques provides promising approach effectively detect irregularities in network traffic, enhancing IoT reducing In this paper, DL-based IDS is proposed using Feed Forward Neural Networks (FFNN), Long Short-Term Memory (LSTM), Random (RandNN) protect networks from cyberattacks. Each DL model its potential benefit as reported paper. For example, FFNN can handle complex traffic patterns, while LSTM good capturing long-term dependencies present traffic. With random connections flexible dynamics, RandNN uses data ability adapt learn data. These algorithms boost cybersecurity by enabling defense mechanisms against challenging threats ensuring sensitive expand. technique exhibits superior performance when compared with current state-of-the-art DL-IDS CIC-IoT22 dataset. An accuracy 99.93 % achieved model, 99.85 96.42 detecting intrusion. Moreover, models have enhance intrusion detection generating swift responses problems

Language: Английский

---

Ensemble Machine Learning of Gradient Boosting (XGBoost, LightGBM, CatBoost) and Attention-Based CNN-LSTM for Harmful Algal Blooms Forecasting

Toxins, Journal Year: 2023, Volume and Issue: 15(10), P. 608 - 608

Published: Oct. 10, 2023

Harmful algal blooms (HABs) are a serious threat to ecosystems and human health. The accurate prediction of HABs is crucial for their proactive preparation management. While mechanism-based numerical modeling, such as the Environmental Fluid Dynamics Code (EFDC), has been widely used in past, recent development machine learning technology with data-based processing capabilities opened up new possibilities prediction. In this study, we developed evaluated two types learning-based models prediction: Gradient Boosting (XGBoost, LightGBM, CatBoost) attention-based CNN-LSTM models. We Bayesian optimization techniques hyperparameter tuning, applied bagging stacking ensemble obtain final results. result was derived by applying optimal techniques, applicability evaluated. When predicting an technique, it judged that overall performance can be improved complementing advantages each model averaging errors overfitting individual Our study highlights potential emphasizes need incorporate latest into important field.

Language: Английский

---

Dugat-LSTM: Deep learning based network intrusion detection system using chaotic optimization strategy

Expert Systems with Applications, Journal Year: 2024, Volume and Issue: 245, P. 123027 - 123027

Published: Jan. 11, 2024

Language: Английский

---

A deep learning-based novel hybrid CNN-LSTM architecture for efficient detection of threats in the IoT ecosystem

Ain Shams Engineering Journal, Journal Year: 2024, Volume and Issue: 15(7), P. 102777 - 102777

Published: April 4, 2024

The Internet of Things (IoT) landscape is witnessing rapid growth, driven by continuous innovation and a simultaneous increase in cybersecurity threats. As these threats become more sophisticated, the imperative to fortify IoT devices against emerging vulnerabilities becomes increasingly pronounced. This research motivated need for comprehensive threat detection solutions that can effectively address evolving landscape. Existing approaches often fall short adapting dynamic nature environments increasing complexity attacks. core problem addressed this development novel Hybrid Convolutional Neural Network Long Short-Term Memory (CNN-LSTM) architecture tailored precise efficient detection. aims overcome limitations existing methods enhance security ecosystems. Our study encompasses detailed analysis proposed CNN-LSTM model, leveraging data from diverse datasets, including IoT-23, N-BaIoT, CICIDS2017. model tested validated on than 14 attack types. We have designed exhibit robust capabilities capturing analyzing data. outcomes our showcase remarkable accuracy, with models achieving 95% accuracy IoT-23 dataset an outstanding 99% both N-BaIoT CICIDS2017 datasets. These findings underscore model's adaptability various environments. contributes significantly enhances introduce Principal Component Analysis (PCA) optimize processing incorporate advanced optimization techniques like quantization pruning improve deployment efficiency resource-constrained lays foundation future advancements bolstering security.

Language: Английский

---

An end-to-end learning approach for enhancing intrusion detection in Industrial-Internet of Things

Knowledge-Based Systems, Journal Year: 2024, Volume and Issue: 294, P. 111785 - 111785

Published: April 10, 2024

Language: Английский

---

Deep Learning Models with Transfer Learning and Ensemble for Enhancing Cybersecurity in IoT Use Cases

International Journal of Computational and Experimental Science and Engineering, Journal Year: 2025, Volume and Issue: 11(1)

Published: Feb. 28, 2025

Internet of Things (IoT) applications have made inroads into different domains, providing unique solutions—Internet technology offers seamless integration physical and digital worlds. However, the broad nature technologies protocols used in IoT has increased vulnerability from malicious attackers. Hence, protecting cyber-attacks is imperative. Researchers implemented intrusion detection systems to overcome this issue improve cybersecurity scenarios. With new threats cybercrime emerging, a continuous effort required enhance security applications. To address pressing need, we present our study that proposes deep learning-based framework bolster at use cases level by exploiting power transfer learning ensembling it models pre-trained larger datasets. Deep attain high performance with help hyperparameter tuning, achieve through PSO proposed system. Our ensemble system shows how individual can outperform using best-performing as constituents approach. We introduce an algorithm called — Optimized Ensemble Learning-Based Intrusion Detection (OEL-ID). This leverages corresponding optimization strategies boost for improved cyber Using UNSW-NB15 benchmark dataset, empirical demonstrates method, compared some existing models, obtained accuracy 98.89%, which, turn, provided highest comparative accuracy. Therefore, be allows significant system's underlying

Language: Английский

---

Res-TranBiLSTM: An intelligent approach for intrusion detection in the Internet of Things

Computer Networks, Journal Year: 2023, Volume and Issue: 235, P. 109982 - 109982

Published: Aug. 12, 2023

The Internet of Things (IoT), as the information carrier and telecommunications networks, is a new network technology comprising physical entities embedded with electronic components, software sensors, characterized by strong complexity openness. With massive amount data, occurrence intrusion also increasingly frequent, involving industrial control systems, IoT devices, mobile security, cloud services, services. diversification intelligence cyberattack behaviors, traditional detection systems (IDSs) face problems—such insufficient feature extraction inaccurate model classification—when faced high-dimensional features nonlinear data. Due to their powerful data representation learning ability, deep methods save substantial time in processing complex On this basis, we propose an using ResNet, Transformer BiLSTM (Res-TranBiLSTM) that takes into account both spatial temporal traffic. We use Synthetic Minor Overriding Technique (SMOTE) – Edited Nearest Neighbor (ENN) method alleviate degree imbalance. In addition, respectively establish based on ResNet extract parallelly. Finally, spatiotemporal are included achieve attack classification. Further, simulation experiments conducted public sets NSL-KDD CIC-IDS2017. implemented Python programming language Pytorch framework. results reveal performance our proposed better than other models, accuracy reaching 90.99%, 99.15% 99.56%, dataset, CIC-IDS2017 dataset MQTTset respectively. It increased about 1%-10% 0.2%-10% dataset. These demonstrate effective constructing optimizing large-scale IDS environment.

Language: Английский

---

Hyperparameter Optimization for 1D-CNN-Based Network Intrusion Detection Using GA and PSO

Mathematics, Journal Year: 2023, Volume and Issue: 11(17), P. 3724 - 3724

Published: Aug. 29, 2023

This study presents a comprehensive exploration of the hyperparameter optimization in one-dimensional (1D) convolutional neural networks (CNNs) for network intrusion detection. The increasing frequency and complexity cyberattacks have prompted an urgent need effective intrusion-detection systems (IDSs). Herein, we focus on optimizing nine hyperparameters within 1D-CNN model, using two well-established evolutionary computation methods—genetic algorithm (GA) particle swarm (PSO). performances these methods are assessed three major datasets—UNSW-NB15, CIC-IDS2017, NSL-KDD. key performance metrics considered this include accuracy, loss, precision, recall, F1-score. results demonstrate considerable improvements all across datasets, both GA- PSO-optimized models, when compared to those original nonoptimized model. For instance, UNSW-NB15 dataset, GA PSO achieve accuracies 99.31 99.28%, respectively. Both algorithms yield equivalent terms Similarly, vary CIC-IDS2017 NSL-KDD indicating that efficacy is context-specific dependent nature dataset. findings importance effects efficient optimization, greatly contributing field security. serves as crucial step toward developing advanced, robust, adaptable IDSs capable addressing evolving landscape cyber threats.

Language: Английский

---

DDoS attacks in Industrial IoT: A survey

Computer Networks, Journal Year: 2023, Volume and Issue: 236, P. 110015 - 110015

Published: Sept. 15, 2023

Language: Английский

---