R. M. Chandrasekaran

Intrusion detection using neural based hybrid classification methods

Data mining is the use of algorithms to extract the information and patterns derived by the knowledge discovery in databases process. Classification is a very common data mining task. Classification maps data into predefined groups or classes. It is often referred to as supervised learning because the classes are determined before examining the data. Due to increasing incidents of cyber attacks, building effective intrusion detection systems are essential for protecting information systems security, and yet it remains an elusive goal and a great challenge. This paper presents two classification methods involving multilayer perceptron and radial basis function and an ensemble of multilayer perceptron and radial basis function. We propose hybrid architecture involving ensemble and base classifiers for intrusion detection systems. The analysis of results shows that the performance of the proposed method is superior to that of single usage of existing classification methods such as multilayer perceptron and radial basis function. Additionally it has been found that ensemble of multilayer perceptron is superior to ensemble of radial basis function classifier for normal behavior and reverse is the case for abnormal behavior. We show that the proposed method provides significant improvement of prediction accuracy in intrusion detection.

Evaluation of k-Nearest Neighbor classifier performance for direct marketing

Text data mining is a process of exploratory data analysis. Classification maps data into predefined groups or classes. It is often referred to as supervised learning because the classes are determined before examining the data. This paper describes the proposed k-Nearest Neighbor classifier that performs comparative cross-validation for the existing k-Nearest Neighbor classifier. The feasibility and the benefits of the proposed approach are demonstrated by means of data mining problem: direct marketing. Direct marketing has become an important application field of data mining. Comparative cross-validation involves estimation of accuracy by either stratified k-fold cross-validation or equivalent repeated random subsampling. While the proposed method may have a high bias; its performance (accuracy estimation in our case) may be poor due to a high variance. Thus the accuracy with the proposed k-Nearest Neighbor classifier was less than that with the existing k-Nearest Neighbor classifier, and the smaller the improvement in runtime the larger the improvement in precision and recall. In our proposed method we have determined the classification accuracy and prediction accuracy where the prediction accuracy is comparatively high.

Role of relay node in Wireless Sensor Network: A survey

This paper describes the concepts of Relay node characteristics, various deployment methods, and their internal behaviors in the Wireless Sensor Networks (WSN). First, the internal behavioral characteristics of relay node and their different state processing are analyzed using algorithm. Then, the influence of relay nodes in WSN and the potential relay node deployment methods were explored and with the efficient deployment of relay nodes, a simple research-level review of WSN accuracy and network energy consumptions are provided via Simulation results. Exposed research issues for the realization of relay node deployment in WSN were also discussed.

A WSN Based Framework for Human Health Monitoring

Wireless Sensor Networks (WSN) is neoteric progressing technology which addresses solutions to critical application environments. It allows the assimilation of medical sensors to form a group scenario which makes it feasible to remotely collect and communicate the physiological signals of a patient, thus preventing the patient from being tied to a particular set of monitoring medical instruments and thereby facilitate the patients prior acquittal from the hospital. The sensor nodes adhered to the human body form a network and transmit the collected data to the medical server at a remote location via internet thus forming a Wireless Body Area Network (WBAN). We propose a framework for WBAN which discusses the working of the WBAN in three phases and the parameters which influence the working of each phase. Further the work also concentrates to generate a full fledged monitoring system for patient monitoring in health care.

Energy and Fault Aware Management Framework for Wireless Sensor Network

Wireless Sensor Networks (WSN) is a collection of numerous tiny sensor nodes which are randomly deployed in distributed environment. The reliability of WSN is affected by faults that may occur due to various reasons. The Fault Tolerance is the key factor for distributed sensor application. The main objective of this paper is to create an algorithm for fault tolerance. According to the network environment changes, this algorithm guarantees reliability. Our paper also gives out a framework which can be a solution for various faults occurring in WSN environment.

Intrusion detection using k-Nearest Neighbor

Data Mining is the use of algorithms to extract the information and patterns derived by the knowledge discovery in databases process. Classification maps data into predefined groups or classes. It is often referred to as supervised learning because the classes are determined before examining the data. In many data mining applications that address classification problems, feature and model selection are considered as key tasks. That is, appropriate input features of the classifier must be selected from a given set of possible features and structure parameters of the classifier must be adapted with respect to these features and a given data set. This paper describes feature selection and model selection simultaneously for k-Nearest Neighbor (k-NN) classifiers. In order to reduce the optimization effort, various techniques are integrated that accelerate and improve the classifier significantly: hybrid k-NN, comparative cross validation. The feasibility and the benefits of the proposed approach are demonstrated by means of data mining problem: intrusion detection in computer networks. It is shown that, compared to earlier k-NN technique, the run time is reduced by up to 0.01 % and 0.06 % while error rates are lowered by up to 0.002 % and 0.03 % for normal and abnormal behaviour respectively. The algorithm is independent of specific applications so that many ideas and solutions can be transferred to other classifier paradigms.

Sentiment Mining Using SVM-Based Hybrid Classification Model

With the rapid growth of social networks, opinions expressed in social networks play an influential role in day-to-day life. A need for a sentiment mining model arises, so as to enable the retrieval of opinions for decision making. Though support vector machine (SVM) has been proved to provide a good classification result in sentiment mining, the practically implemented SVM is often far from the theoretically expected level because their implementations are based on the approximated algorithms due to the high complexity of time and space. To improve the limited classification performance of the real SVM, we propose to use the hybrid model of SVM and principal component analysis (PCA). In this paper, we apply the concept of reducing the data dimensionality using PCA to decrease the complexity of an SVM-based sentiment classification task. The experimental results for the product reviews show that the proposed hybrid model of SVM with PCA outperforms a single SVM in terms of classification accuracy and receiver-operating characteristic curve (ROC).

Signature verification using radial basis function classifier

Data Mining is the use of algorithms to extract the information and patterns derived by the knowledge discovery in databases process. Classification maps data into predefined groups or classes. It is often referred to as supervised learning because the classes are determined before examining the data. The Verification of handwritten Signature, which is a behavioral biometric, can be classified into off-line and online signature verification methods. The feasibility and the benefits of the proposed approach are demonstrated by means of data mining problem: online Signature Verification This paper addresses using ensemble approach of Radial Basis Function Classifier for online Signature Verification. Online signature verification, in general, gives a higher verification rate than off-line verification methods, because of its use of both static and dynamic features of problem space in contrast to off-line which uses only the static features. We show that proposed ensemble of Radial Basis Function classifier is superior to individual approach for Signature Verification in terms of classification rate.

A Cross Layer Approach for Fault Management Framework in Wireless Sensor Networks

Wireless sensor network (WSN) is a collection of sensor nodes which are randomly deployed in distributed environment. Fault detection and management is major criteria in wireless sensor networks. Cross layer approach is used to implement Fault management plane in wireless sensor networks. Cross-layer approach (XLA) is developed, which replaces the traditional layered approach (TLA) that has been used in Wireless sensor networks. The design principle of XLA is both the data and the functional operations of traditional communication layers are melted in a single protocol. The objective of this paper is creating a cross layer approach (XLA) algorithm for fault management and it is verified using high level language.

Effect of Principle Component Analysis and Support Vector Machine in Software Fault Prediction

Machine Learning (ML) approaches have a great impact in fault prediction. Demand for producing quality assured software in an organization has been rapidly increased during the last few years. This leads to increase in development of machine learning algorithms for analyzing and classifying the data sets, which can be used in constructing models for predicting the important quality attributes such as fault proneness. Defective modules in software project have a considerable risk which reduces the quality of the software. This paper mainly addresses the software fault prediction using hybrid Support Vector Machine (SVM) classifier. We conduct a comparative study using the WEKA tool for three different levels of software metrics (package level, class level and method level) with hybrid SVM classifiers using feature selection techniques such as Principle Component Analysis (PCA). The experiments are carried out on the datasets such as NASA KC1 method level data set, NASA KC1 class level dataset and Eclipse dataset for package level metrics. The feature selection techniques evolved by experiments shows that Principle Component Analysis (PCA) with hybrid SVM performs better than other feature selection techniques.