A SURVEY ON DIFFERENT MACHINE LEARNING ALGORITHMS AND WEAK CLASSIFIERS BASED ON KDD AND NSL-KDD DATASETS

Rama Devi Ravipati,Munther Abualkibash

doi:10.5121/ijaia.2019.10301

Abstract

Network intrusion detection often finds a difficulty in creating classifiers that could handle unequal distributed attack categories. Generally, attacks such as Remote to Local (R2L) and User to Root (U2R) attacks are very rare attacks and even in KDD dataset, these attacks are only 2% of overall datasets. So, these result in model not able to efficiently learn the characteristics of rare categories and this will result in poor detection rates of rare attack categories like R2L and U2R attacks. We even compared the accuracy of KDD and NSL-KDD datasets using different classifiers in WEKA.

Highlights

On a computing platform Intrusion is a set of actions that attempts to compromise the integrity, confidentiality, or availability of any resource
Many researchers have used intrusion detection system based on supervised learning approach, such as neural network (NN)-based approach and support vector machine (SVM)-based approach
Deep learning is a part of machine learning that is classified based on the layers present in the artificial neural network

Summary

INTRODUCTION

On a computing platform Intrusion is a set of actions that attempts to compromise the integrity, confidentiality, or availability of any resource. Network based intrusion prevention system – It analyses the protocol activity by monitoring the entire network for suspicious traffic. B. Wireless Intrusion prevention system – It analyses the wireless networking protocol by monitoring the entire network for suspicious traffic. A. Signature-based IDS refers to the attacks in the network traffic by looking for specific patterns, such as byte sequences [1]. Signature-based IDS refers to the attacks in the network traffic by looking for specific patterns, such as byte sequences [1] It only detects the known attacks but not new attacks because no pattern is available for those new attacks [2]. It uses machine learning to create a model and compare it with the new behaviour This approach helps to detect the new attacks, but it suffers from false positives. An Unsupervised learning algorithm finds the hidden pattern in the data, whereas a supervised learning algorithm finds the relationship between data and its class

RELATED WORK

Probing

OVERVIEW OF OUR ALGORITHM

Update the weight for each datapoint as:

For all output neurons i do

Parameters used for classification

Precision

Nearest Neighbour

Findings

CONCLUSION AND FUTURE WORK