An AdamW-Based Deep Neural Network Using Feature Selection and Data Oversampling for Intrusion Detection DOI

Zhuoer Lu,

Xiaoyong Li, Pengfei Qiu

et al.

Published: Aug. 18, 2023

With the development of Internet and increasing number users, cyber security has become a major concern for most netizens. In this paper, we propose an AdamW-based neural network using feature selection data oversampling intrusion detection. First, use Random Forest classifier to select 25 important features classifying traffic. Second, given imbalance different types samples in NSL-KDD dataset, ADASYN oversample minority samples. addition, achieve better performance, AdamW as optimizer our deep network. Finally, tune hyperparameters get best classification results Compared with other classical machine learning models detection, achieves high detection performance: test set loss is reduced 0.0001 accuracy improved 99.8%.

Language: Английский

Crested Porcupine Optimizer: A new nature-inspired metaheuristic DOI
Mohamed Abdel‐Basset, Reda Mohamed, Mohamed Abouhawwash

et al.

Knowledge-Based Systems, Journal Year: 2023, Volume and Issue: 284, P. 111257 - 111257

Published: Dec. 22, 2023

Language: Английский

Citations

146
Multi-strategy RIME optimization algorithm for feature selection of network intrusion detection DOI
Lan Wang,

Jialing Xu,

Liyun Jia

et al.

Computers & Security, Journal Year: 2025, Volume and Issue: unknown, P. 104393 - 104393

Published: Feb. 1, 2025

Language: Английский

Citations

0
An enhanced moth flame optimization extreme learning machines hybrid model for predicting CO2 emissions DOI Creative Commons

Ahmed Ramdan Almaqtouf Algwil,

Wagdi Khalifa

Scientific Reports, Journal Year: 2025, Volume and Issue: 15(1)

Published: April 8, 2025

Language: Английский

Citations

0
Optimized Intrusion Detection Approach for Cyber‐Physical System Using Meta‐Learning With Stacked Generalization: An Ensemble Learning Inspired Approach DOI
Ram Ji, Neerendra Kumar,

Devanand Padha

et al.

Security and Privacy, Journal Year: 2025, Volume and Issue: 8(3)

Published: April 27, 2025

ABSTRACT Cyber‐physical systems (CPSs) are crucial in providing vital infrastructure like smart grids, cities, automobiles, healthcare systems, and so forth, for many nations. CPSs vulnerable to various attacks due their large attack surface. An on these may lead the disruption of critical services. To protect an optimized intrusion detection approach is needed. Although approaches exist, they have limitations poor accuracy, high time, space time complexities, false alarm rates, etc. stack generalized meta‐learner‐based has been proposed this paper. The utilizes numerous core models a meta‐learner classify network traffic CPSs. base trained learning data, outcomes used as input features meta‐learner, which then makes final prediction. Four classifiers being models, namely random forest (RF), gradient boosting (GB), multiple layer perceptron (MLP), k ‐nearest neighbors (KNNs), extreme (XGB) classifier meta‐learner. predictions generated using stacking ensemble approach. Auto encoders feature extraction, thereby utilizing unique objective function designed recursive attribute elimination. presented selects only 10 out 46 features, helps reducing complexities. While implementing CIC‐IoT‐2023 dataset, following results obtained: multi‐classification accuracy (98.94%), precision (0.99), recall F 1 score average positive rate (0.0003), (0.12 s). When implemented NSL‐KDD (99%), (0.0012). UNSW‐NB15 (99.56%), (0.0002). performs better contrast other cutting‐edge approaches. Also, introduces novel effective strategy

Language: Английский

Citations

0
Advancing feature ranking with hybrid feature ranking weighted majority model: a weighted majority voting strategy enhanced by the Harris hawks optimizer DOI Creative Commons
Mansourah Aljohani, Yousry AbdulAzeem, Hossam Magdy Balaha

et al.

Journal of Computational Design and Engineering, Journal Year: 2024, Volume and Issue: 11(3), P. 308 - 325

Published: May 1, 2024

Abstract Feature selection (FS) is vital in improving the performance of machine learning (ML) algorithms. Despite its importance, identifying most important features remains challenging, highlighting need for advanced optimization techniques. In this study, we propose a novel hybrid feature ranking technique called Hybrid Ranking Weighted Majority Model (HFRWM2). HFRWM2 combines ML models with Harris Hawks Optimizer (HHO) metaheuristic. HHO known versatility addressing various challenges, thanks to ability handle continuous, discrete, and combinatorial problems. It achieves balance between exploration exploitation by mimicking cooperative hunting behavior Harris’s hawks, thus thoroughly exploring search space converging toward optimal solutions. Our approach operates two phases. First, an odd number models, conjunction HHO, generate encodings along metrics. These are then weighted based on their metrics vertically aggregated. This process produces rankings, facilitating extraction top-K features. The motivation behind our research 2-fold: enhance precision algorithms through optimized FS improve overall efficiency predictive models. To evaluate effectiveness HFRWM2, conducted rigorous tests datasets: “Australian” “Fertility.” findings demonstrate navigating We compared 12 other techniques found it outperform them. superiority was particularly evident graphical comparison dataset, where showed significant advancements ranking.

Language: Английский

Citations

2
An Adaptation of Hybrid Binary Optimization Algorithms for Medical Image Feature Selection in Neural Network for Classification of Breast Cancer DOI
Olaide N. Oyelade, Enesi Femi Aminu, Hui Wang

et al.

Neurocomputing, Journal Year: 2024, Volume and Issue: unknown, P. 129018 - 129018

Published: Nov. 1, 2024

Language: Английский

Citations

2
Nonlinear optimization of optical camera multiparameter via triple integrated Gradient-based optimizer algorithm DOI
Kangjian Sun, Ju Huo, Heming Jia

et al.

Optics & Laser Technology, Journal Year: 2024, Volume and Issue: 179, P. 111294 - 111294

Published: June 15, 2024

Language: Английский

Citations

1
An ensemble system for machine learning IoT intrusion detection based on enhanced artificial hummingbird algorithm DOI
Leyi Shi, Qihang Yang, Lin Gao

et al.

The Journal of Supercomputing, Journal Year: 2024, Volume and Issue: 81(1)

Published: Nov. 1, 2024

Language: Английский

Citations

1
Utilizing bee foraging behavior in mutational salp swarm for feature selection: a study on return-intentions of overseas Chinese after COVID-19 DOI Creative Commons
Jie Xing,

Qinqin Zhao,

Huiling Chen

et al.

Journal of Computational Design and Engineering, Journal Year: 2023, Volume and Issue: 10(6), P. 2094 - 2121

Published: Oct. 19, 2023

Abstract We present a bee foraging behavior-driven mutational salp swarm algorithm (BMSSA) based on an improved strategy and unscented mutation strategy. The is leveraged in the follower location update phase to break fixed range search of algorithm, while optimal solution employed enhance quality solution. Extensive experimental results public CEC 2014 benchmark functions validate that proposed BMSSA performs better than nine well-known metaheuristic methods seven state-of-the-art algorithms. binary (bBMSSA) further for feature selection by using as support vector machine classifier. Experimental comparisons 12 UCI datasets demonstrate superiority bBMSSA. Finally, we collected dataset return-intentions overseas Chinese after coronavirus disease (COVID-19) through anonymous online questionnaire performed case study setting up bBMSSA-based optimization model. outcomes manifest model exhibits conspicuous prowess, attaining accuracy exceeding 93%. shows development prospects, family job place residence, seeking opportunities China, possible time return China are critical factors influencing willingness COVID-19.

Language: Английский

Citations

2
Reinforcement learning guided Spearman dynamic opposite Gradient-based optimizer for numerical optimization and anchor clustering DOI Creative Commons
Kangjian Sun,

Ju Huo,

Heming Jia

et al.

Journal of Computational Design and Engineering, Journal Year: 2023, Volume and Issue: 11(1), P. 12 - 33

Published: Dec. 20, 2023

Abstract As science and technology advance, the need for novel optimization techniques has led to an increase. The recently proposed metaheuristic algorithm, Gradient-based optimizer (GBO), is rooted in gradient-based Newton's method. GBO a more concrete theoretical foundation. However, gradient search rule (GSR) local escaping operator (LEO) operators still have some shortcomings. insufficient updating method simple selection process limit performance of algorithm. In this paper, improved version compensate above shortcomings, called RL-SDOGBO. First, during GSR phase, Spearman rank correlation coefficient used determine weak solutions on which perform dynamic opposite learning. This operation assists algorithm escape from optima enhance exploration capability. Secondly, optimize exploitation capability, reinforcement learning guide solution update modes LEO operator. RL-SDOGBO tested 12 classical benchmark functions CEC2022 with seven representative metaheuristics, respectively. impact improvements, scalability running time balance are analyzed discussed. Combining experimental results statistical results, exhibits excellent numerical provides high-quality most cases. addition, also solve anchor clustering problem small target detection, making it potential competitive option.

Language: Английский

Citations

2