Effective Intrusion Detection System Using XGBoost

Dhaliwal, Sukhpreet Singh; Nahid, Abdullah-Al; Abbas, Robert

doi:10.3390/info9070149

Cited by 297 publications

(138 citation statements)

References 17 publications

Supporting

Mentioning

133

Contrasting

Unclassified

Order By: Relevance

“…gradient boosting models, its implementation of parallel processing enables a fast model training compared to many traditional models, and can be deployed to high-performance platform for large-scale parallel computing. The technique was found to outperform other machine learning and deep learning techniques in many competitions such as Kaggle and KDDCup (Chen and Guestrin, 2016;Dhaliwal, et al, 2018), especially for datasets with sparse matrix. It has been successfully applied in many bioinformatic studies, such as miRNA-disease association (Chen, et al, 2018), protein translocation (Mendik, et al, 2019), protein-protein interactions (Basit, et al, 2018), and DNA methylation (Zou, et al, 2018).…”

mentioning

confidence: 98%

Accurate Prediction of Genome-wide RNA Secondary Structure Profile Based On Extreme Gradient Boosting

Rao

Zhao

et al. 2019

Preprint

View full text Add to dashboard Cite

Motivation:Many studies have shown that RNA secondary structure plays a vital role in fundamental cellular processes, such as protein synthesis, mRNA processing, mRNA assembly, ribosome function and eukaryotic spliceosomes. Identification of RNA secondary structure is a key step to understand the common mechanisms underlying the translation process. Recently, a few experimental methods were developed to measure genomewide RNA secondary structure profile through high-throughput sequencing techniques, and have been successfully applied to genomes including yeast and human. However, these high-throughput methods usually have low precision and are hard to cover all nucleotides on the RNA due to limited sequencing coverage. Results:In this study, we developed a new method for the prediction of genome-wide RNA secondary structure profile (TH-GRASP) from RNA sequence based on eXtreme Gradient Boosting (XGBoost). The method achieves an prediction with areas under the receiver operating characteristic curve (AUC) values greater than 0.9 on three different datasets, and AUC of 0.892 by an independent test on the recently released Zika virus RNA dataset. These AUCs represent a consistent increase of >6% than the recently developed method CROSS trained by a shallow neural network. A further analysis on the 1000-Genome Project data showed that our predicted unpaired probability at mutations sites are highly correlated with the minor allele frequencies (MAF) of synonymous, nonsynonymous mutations, and mutations in 3' and 5'UTR with Pearson Correlation Coefficients all above 0.8. These PCCs are consistently higher than those generated by RNAplfold method. Moreover, an investigation over all human mRNA indicated a periodic distribution of the predicted unpaired probability on codons, and a decrease of paired probability in the boundary with 5' and 3' untranslated regions. These results highlighted TH-GRASP is effective to remove experimental noises and to have ability to make predictions on nucleotides with low or no coverage by fitting high-throughput genomic data for RNA secondary structure profiles, and also suggested that building model on high throughput experimental data might be a future direction to substitute analytical methods.

show abstract

mentioning

confidence: 98%

Accurate Prediction of Genome-wide RNA Secondary Structure Profile Based On Extreme Gradient Boosting

Rao

Zhao

et al. 2019

Preprint

View full text Add to dashboard Cite

show abstract

“…Classification models were built using a powerful tree boosting algorithm, XGBoost, which in 2015 was used in every winning team in the top 10 of the Data Mining and Knowledge Discovery competition for a wide range of machine learning problems [19] and was suggested to be one of the most sophisticated methods at the time of this work [20], [21]. Furthermore, the tree learning algorithm uses parallel and distributed computing and is approximately 10 times faster than existing methods and allows many hyperparameters to be tuned to reduce the chance of overfitting [20].…”

Section: Methodsmentioning

confidence: 99%

Working Towards a Blood-Derived Gene Expression Biomarker Specific for Alzheimer’s Disease

Patel

Iniesta

Ståhl

et al. 2019

Preprint

View full text Add to dashboard Cite

BackgroundA significant number of studies have investigated the use of blood-derived gene expression profiling as a biomarker for Alzheimer's Disease (AD). However, the typical approach of developing classification models trained on subjects with AD and complimentary cognitive healthy controls may result in markers of general illness rather than being AD-specific.Incorporating additional related neurological and age-related disorders during the classification model development process may lead to the discovery of an AD-specific expression signature. MethodsTwo XGBoost classification models were developed and optimised. The first used the typical approach, training on 160 AD and 160 cognitively normal controls, while the second was trained in 6318 AD and 6318 mixed controls. Up-sampling was performed in each training set to the minority classes to avoid sampling bias, and both classification models were evaluated in an independent dataset consisting of 127 AD and 687 mixed controls. The mixed control group represents a heterogeneous ageing population consisting of Parkinson's Disease, Multiple Sclerosis, Amyotrophic Lateral Sclerosis, Bipolar Disorder, Schizophrenia, Coronary Artery Disease, Rheumatoid Arthritis, Chronic Obstructive Pulmonary Disease, and cognitively healthy subjects. ResultsThe typical approach resulted in a 74 gene classification model with a validation performance of 58.3% sensitivity, 30.3% specificity, 13.4% PPV and 79.7% NPV. In contrast, the second approach resulted in a 28 gene classification model with an overall improved validation performance of 46.5% sensitivity, 95.6% specificity, 66.3% PPV and 90.6% NPV. ConclusionsThe addition of related neurological and age-related disorders into the AD classification model developmental process identified a more AD-specific expression signature, with improved ability to distinguish AD from other related diseases and cognitively healthy controls. However, this was at the cost of sensitivity. Further improvement is still required to identify a robust blood transcriptomic signature specific to AD.

show abstract

“…Various methods have been proposed in the literature for network anomaly detection including standard machine learning classifiers 4–29 and deep learning techniques 30–47 . Muda et al performed clustering before classification and compared the single classifiers with hybrid classifiers.…”

Section: Related Workmentioning

confidence: 99%

“…Mirza developed an ensemble method that combines logistic regression, neural networks, and decision trees and obtained 96.14% detection rate using KDDCup99 dataset 25 . Dhaliwal et al developed several XGBoost models and obtained 98.70% accuracy on NSL‐KDD dataset 26 . Dahiya and Srivastava compared two dimension reduction algorithms such as canonical correlation analysis and linear discriminant analysis using several classification algorithms and obtained at most 95.53% accuracy rate on UNSW‐NB dataset using canonical correlation analysis with bagging 27 .…”

Section: Related Workmentioning

confidence: 99%

A deep learning approach with Bayesian optimization and ensemble classifiers for detecting denial of service attacks

Görmez

Aydın

Karademir³

et al. 2020

Int J Communication

View full text Add to dashboard Cite

Summary Detecting malicious behavior is important for preventing security threats in a computer network. Denial of Service (DoS) is among the popular cyber attacks targeted at web sites of high‐profile organizations and can potentially have high economic and time costs. In this paper, several machine learning methods including ensemble models and autoencoder‐based deep learning classifiers are compared and tuned using Bayesian optimization. The autoencoder framework enables to extract new features by mapping the original input to a new space. The methods are trained and tested both for binary and multi‐class classification on Digiturk and Labris datasets, which were introduced recently for detecting various types of DDoS attacks. The best performing methods are found to be ensembles though deep learning classifiers achieved comparable level of accuracy.

show abstract

Effective Intrusion Detection System Using XGBoost

Cited by 297 publications

References 17 publications

Accurate Prediction of Genome-wide RNA Secondary Structure Profile Based On Extreme Gradient Boosting

Accurate Prediction of Genome-wide RNA Secondary Structure Profile Based On Extreme Gradient Boosting

Working Towards a Blood-Derived Gene Expression Biomarker Specific for Alzheimer’s Disease

A deep learning approach with Bayesian optimization and ensemble classifiers for detecting denial of service attacks

Contact Info

Product

Resources

About