An Approach for Optimizing Ensemble Intrusion Detection Systems

Stiawan, Deris; Heryanto, Ahmad; Bardadi, Ali; Rini, Dian Palupi; Subroto, Imam Much Ibnu; Kurniabudi, Kurniabudi; Idris, Mohd. Yazid; Abdullah, Abdul Hanan; Kerim, Bedine; Budiarto, Rahmat

doi:10.1109/access.2020.3046246

Cited by 25 publications

(12 citation statements)

References 59 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This algorithm is used in cyber-physical social networksReview on classifiers: Classifiers includes Naïve Bayes, Random Forest [19], J-48 and bagging [23]are widely used data mining classifiers to classify a class.Random_Forest: categorizes all the samples in the trained dataset and produces good results but it is over-fit to probability predictions.Naïve Bayes: is a simple supervised procedure that returns likelihoods. j48 [27]:This algorithm generates the decision tree for the taken dataset. The accuracy level obtained using J48 [26] is best compared to the Naive Bayes classifier.…”

Section: Materials and Methodologiesmentioning

confidence: 99%

RESMOTE: A New Approach for Class Imbalance Problem

Sasirekha

Kanisha

2023

Advances in Transdisciplinary Engineering

View full text Add to dashboard Cite

Data mining is a growing domain. Wherever we are going, data mining Techniques are available to collect important information from the warehouse or from databases. In every sector, they need to keep data secure. Maintaining a balanced state of data is very crucial. Whenever a class tends to classify, the specimens in the class can be grouped as the majority group and the minority group. The majority group consists of higher number of data when compared with the data distributed in the minority group. This paper proposes the Randomized-Ensemble with Smote (RESMOTE) methodology to handle the class imbalance problem. This paper addresses imbalance issues and proposes a class imbalance solution using a sampling technique combined with certain classification metrics.

show abstract

Section: Materials and Methodologiesmentioning

confidence: 99%

RESMOTE: A New Approach for Class Imbalance Problem

Sasirekha

Kanisha

2023

Advances in Transdisciplinary Engineering

View full text Add to dashboard Cite

show abstract

“…A clustering algorithm evaluates posts suspected of containing inhumane words. D. Stiawan et al, (2021), the author proposes a new dataset IDS that can be used to identify the best-fitting selected features as critical features. A method for developing an optimally integrated IDS to achieve this goal is developedIt is used to select six parameters namely Information Gain (IG), Gain Ratio (GR), Symmetric Uncertainty (SU), Relief-F (R-F), One-R (OR) and Chi-Square (CS).…”

Section: Our Contributions Can Be Summarized As Followsmentioning

confidence: 99%

Deep Learning based Densenet Convolution Neural Network for Community Detection in Online Social Networks

Selvakumar,

Kathiravan

2023

IJRITCC

View full text Add to dashboard Cite

Online Social Networks (OSNs) have become increasingly popular, with hundreds of millions of users in recent years. A community in a social network is a virtual group with shared interests and activities that they want to communicate. OSN and the growing number of users have also increased the need for communities. Community structure is an important topological property of OSN and plays an essential role in various dynamic processes, including the diffusion of information within the network. All networks have a community format, and one of the most continually addressed research issues is the finding of communities. However, traditional techniques didn't do a better community of discovering user interests. As a result, these methods cannot detect active communities. To tackle this issues, in this paper presents Densenet Convolution Neural Network (DnetCNN) approach for community detection. Initially, we gather dataset from Kaggle repository. Then preprocessing the dataset to remove inconsistent and missing values. In addition to User Behavior Impact Rate (UBIR) technique to identify the user URL access, key term and page access. After that, Web Crawling Prone Factor Rate (WCPFR) technique is used find the malicious activity random forest and decision method. Furthermore, Spider Web Cluster Community based Feature Selection (SWC2FS) algorithm is used to choose finest attributes in the dataset. Based on the attributes, to find the community group using Densenet Convolution Neural Network (DnetCNN) approach. Thus, the experimental result produce better performance than other methods.

show abstract

“…The best models were Random Forest with an accuracy of 99.86% and 22 features and J48 with an accuracy of 99.87% and 52 features. D. Stiawan et al [60] introduced an approach for constructing ensemble IDS using six ranked feature selection techniques, namely, Information Gain, Gain Ratio, Symmetrical Uncertainty, Relief-F, One-R and Chi-Square ensemble with four classifiers such as Bayesian Network, Naïve Bayesian, J48 and SOM, and validated using Hold-up, K-fold approaches. Experimental results were obtained on Weka using the ITD-UTM dataset.…”

Section: Machine Learning-based Idsmentioning

confidence: 99%

Importance of Machine Learning Techniques to Improve the Open Source Intrusion Detection Systems

Vadhil

Nanne

Salihi

2021

IJEEI

View full text Add to dashboard Cite

Nowadays, it became difficult to ensure data security because of the rapid development of information technology according to the Vs of Big Data. To secure a network against malicious activities and to ensure data protection, an intrusion detection system played a very important role. The main objective was to obtain a high-performance solution capable of detecting different types of attacks around the system. The main aim of this paper is to study the lacks of traditional and open source Intrusion Detection Systems and the Machine Learning techniques commonly used to overcome these lacks. A comparison of some existing works by Intrusion Detection System type, detection method, algorithm and accuracy was provided.

show abstract

An Approach for Optimizing Ensemble Intrusion Detection Systems

Cited by 25 publications

References 59 publications

RESMOTE: A New Approach for Class Imbalance Problem

RESMOTE: A New Approach for Class Imbalance Problem

Deep Learning based Densenet Convolution Neural Network for Community Detection in Online Social Networks

Importance of Machine Learning Techniques to Improve the Open Source Intrusion Detection Systems

Contact Info

Product

Resources

About