Hybrid Deep Learning Based Attack Detection for Imbalanced Data Classification

Almarshdi, Rasha; Nassef, Laila; Fadel, Etimad; Alowidi, Nahed

doi:10.32604/iasc.2023.026799

Cited by 16 publications

(5 citation statements)

References 33 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…They used SMOTE to enhance model performance on the UNSW-NB15, KDD99, and NSL-KDD datasets, achieving accuracy scores of 91.2%, 95.2%, and 82.6%, respectively. e) Almarshdi et al [29] developed a hybrid deep learning IDS using convolution neural network (CNN) and long short-term memory (LSTM) algorithms, combined with the SMOTE technique. This model achieved 92.10% accuracy on the UNSW-NB15 dataset compared to 89.90% for the basic CNN model.…”

Section: Common Strategies For Addressing Imbalanced Datamentioning

confidence: 99%

Addressing Imbalanced Data in Network Intrusion Detection: A Review and Survey

Al-Qarni,

Al-Asmari

2024

IJACSA

View full text Add to dashboard Cite

The proliferation of internet-connected devices, including smartphones, smartwatches, and computers, has led to an unprecedented surge in data generation. The rapid rise in device connectivity points to an urgent need for robust cybersecurity measures to counter the mounting wave of cyber threats. Among the strategies aimed at establishing efficient network intrusion detection systems, the integration of machine learning techniques is a prominent avenue. However, the application of machine learning models to imbalanced intrusion detection datasets, such as NSL-KDD, CICIDS2017, and UGR'16, presents challenges. In such intricate scenarios, accurately distinguishing network intrusions poses a formidable challenge. The term "imbalance" refers to the imbalanced distribution of data across classes, which adversely affects the precision of machine learning algorithm classifications. This comprehensive survey embarks on a thorough exploration of the spectrum of methodologies proposed to address the challenge of imbalanced data. Simultaneously, it assesses the efficacy of these methodologies within the realm of network intrusion detection. Moreover, by shedding light on the potential consequences of not effectively tackling imbalanced data, this study aims to provide a holistic understanding of the intricate interplay between machine learning and intrusion detection in imbalanced settings.

show abstract

Section: Common Strategies For Addressing Imbalanced Datamentioning

confidence: 99%

Addressing Imbalanced Data in Network Intrusion Detection: A Review and Survey

Al-Qarni,

Al-Asmari

2024

IJACSA

View full text Add to dashboard Cite

show abstract

“…Each convolutional layer is followed by a pooling layer with a pooling size of 2 and Max-pooling methods. The formula of the max-pooling layer is calculated as [41]:…”

Section: Endmentioning

confidence: 99%

Ensemble Voting-Based Anomaly Detection for a Smart Grid Communication Infrastructure

Alshede¹,

Nassef²,

Alowidi³

et al. 2023

Intelligent Automation &Amp; Soft Computing

Self Cite

View full text Add to dashboard Cite

Advanced Metering Infrastructure (AMI) is the metering network of the smart grid that enables bidirectional communications between each consumer's premises and the provider's control center. The massive amount of data collected supports the real-time decision-making required for diverse applications. The communication infrastructure relies on different network types, including the Internet. This makes the infrastructure vulnerable to various attacks, which could compromise security or have devastating effects. However, traditional machine learning solutions cannot adapt to the increasing complexity and diversity of attacks. The objective of this paper is to develop an Anomaly Detection System (ADS) based on deep learning using the CIC-IDS2017 dataset. However, this dataset is highly imbalanced; thus, a two-step sampling technique: random under-sampling and the Synthetic Minority Oversampling Technique (SMOTE), is proposed to balance the dataset. The proposed system utilizes a multiple hidden layer Auto-encoder (AE) for feature extraction and dimensional reduction. In addition, an ensemble voting based on both Random Forest (RF) and Convolutional Neural Network (CNN) is developed to classify the multiclass attack categories. The proposed system is evaluated and compared with six different state-ofthe-art machine learning and deep learning algorithms: Random Forest (RF), Light Gradient Boosting Machine (LightGBM), eXtreme Gradient Boosting (XGboost), Convolutional Neural Network (CNN), Long Short-Term Memory (LSTM), and bidirectional LSTM (biLSTM). Experimental results show that the proposed model enhances the detection for each attack class compared with the other machine learning and deep learning models with overall accuracy (98.29%), precision (99%), recall (98%), F 1 score (98%), and the UNDetection rate (UND) (8%).

show abstract

“…Since then, the LSTM has grown to be one of the most widely used models in the artificial intelligence field and has been used in many other applications, including sentiment analysis, machine translation, and handwriting recognition. Several previous studies have disclosed LSTM's superior ability to predict related data with temporal properties, such as time series data and text [3], [4].…”

Section: Introductionmentioning

confidence: 99%

Long Short-Term Memory (LSTM): Trends and Future Research Potential

Arifin¹,

Wijaya²,

Nariswari³

et al. 2023

IJETAE

View full text Add to dashboard Cite

-One of the most widely used machine learning methods, Long Short-Term Memory (LSTM), is particularly useful for time series prediction. In this study, we carried out a bibliometric analysis against publications about LSTMs to identify trends and contributions of researchers in the development of machine learning technology. We collect bibliometric data from the Scopus database and use the bibliometric analysis method to analyze trends and contributions of researchers in publications about LSTM. Results of the bibliometric analysis show that LSTM is a lot used in related machine learning applications with time series data and is one the most popular technique for use in predictions. In addition, the use of LSTM is often combined with other deep learning methods, such as neural networks, to improve accuracy prediction. In addition, the results of the bibliometric analysis also show that the use of LSTM has spread to various fields, such as in handwriting recognition, processing Language experience, and recognition of a face. Implications from the results of this study are that the use of LSTM can provide solutions that are accurate and effective in solving prediction problems in various fields, especially in practical applications such as business, health, and transportation. The results of the LSTM bibliometric analysis can provide a broader view of trends and the contributions of researchers to the development of machine learning technology, as well as identify potential research areas for further development. Therefore, this research provides an important contribution to strengthening the results of previous research and showing that the use of LSTM has great potential in the development of future machine learning technology

show abstract

Hybrid Deep Learning Based Attack Detection for Imbalanced Data Classification

Cited by 16 publications

References 33 publications

Addressing Imbalanced Data in Network Intrusion Detection: A Review and Survey

Addressing Imbalanced Data in Network Intrusion Detection: A Review and Survey

Ensemble Voting-Based Anomaly Detection for a Smart Grid Communication Infrastructure

Long Short-Term Memory (LSTM): Trends and Future Research Potential

Contact Info

Product

Resources

About