Volume 4, Issue 8, August 2012

Comparison of the C4.5 and a Naive Bayes Classifier for the Prediction of Lung Cancer Survivability

George Dimitoglou, James A. Adams, and Carol M. Jim

Numerous datamining techniques have been developed to extract information and identify patterns and predict trends from large data sets. In this study, two classification techniques, the J48 implementation of the C4.5 algorithm and a Naive Bayes classifier are applied to predict lung cancer survivability from an extensive data set with fifteen years of patient records. The purpose of the project is to verify the predictive effectiveness of the two techniques on real, historical data. Besides the performance outcome that renders J48 marginally better than the Naive Bayes technique, there is a detailed description of the data and the required pre-processing activities. The performance results confirm expectations while some of the issues that appeared during experimentation, under score the value of having domain-specific understanding to leverage any domain-specific characteristics inherent in the data.

Keywords: Data mining, mining methods and algorithms, text mining

Facial Expression Recognition Based on Moment Invariants

Renuka R. Londhe and Vrushsen P. Pawar

Facial Expression Recognition is rapidly becoming area of interest in computer science and human computer interaction because the most expressive way of displaying the emotions by human is through the facial expressions. In this paper, recognition of facial expression is studied with the help of several properties associated with the face itself. As facial expression changes, the curvatures on the face and properties of the objects such as eyebrows, nose, lips and mouth area also changes. We have used Hu Moment invariants to compute these changes and computed results (changes) are recorded as feature vectors. We have introduced a method for facial expression recognition using Hu Moment Invariants as features. We have used Artificial Neural Network as a classification tool and we developed associated scheme. The Generalized Feedforward Neural Network recognizes six universal expressions i.e. anger, disgust, fear, happy, sad, and surprise as well as seventh one neutral. The Neural Network trained and tested by using Scaled Conjugate Gradient Backpropogation Algorithm. As a result we got 92.4 % classification rate with testing performance 0.0130.

Keywords: Artificial Neural Network, Facial Expressions, Hu Moment Invariant, Human Computer Interaction

Investigating the Statistical Linear Relation between the Model Selection Criterion and the Complexities of Data Mining Algorithms

Dost Muhammad Khan, Nawaz Mohamudally and D K R Babajee

The “model selection criterion” plays a vital role in the preparation of the readiness of a dataset for further data mining actions. It is a gauge to determine whether the dataset is under-fitted or over-fitted. In such cases the datasets are not suitable for knowledge extraction; otherwise the knowledge obtained would be vague, ambiguous and might be misleading. In this paper, we investigate the linear relation between the selection of data mining algorithms through the model selection criterion, the Akaike Information Criterion (AIC) or the Bayesian Information Criterion (BIC), the two most commonly used methodologies in an attempt to set up the dataset for the data mining process cycle. Moreover the complexities of data mining algorithms together with the AIC or BIC at different steps within the data mining process cycle are evaluated to apply the best algorithm in view of generating the optimum accuracy of the knowledge.

Keywords: AIC, BIC, Over-fitted, Under-fitted, Model Selection Criterion, Linear Model, Correlation

The Formulation of a Data Mining Theory for the Knowledge Extraction by means of a Multiagent System

Dost Muhammad Khan, Nawaz Mohamudally and D K R Babajee

Data mining is an extraction of useful information from datasets which provides the knowledge to ensure the effective decision making. The extraction of the knowledge from the dataset is not a single-step process rather then it is a multi-step process which involves the different data mining processes such as clustering, classification and visualization, where the output of one data mining process is the input of another process. The existing data mining algorithms and techniques only focus on one specific data mining task at a time, thus fail to address the fundamental need of supporting the extraction of knowledge as a multi-step process. Therefore, in this paper we propose a multi-step data mining theory (DMT) and a framework for the extraction of knowledge called knowledge extraction process (KEP). The clusters of the given dataset are generated in the first step; the second step consists of two data mining processes classification and visualization which takes the inputs of first step and generate the ‘knowledge’ as final output. A MAS approach is used to validate the proposed DMT.

Keywords: MAS, DMT, KDD, KEP

RNS Overflow Detection Scheme for the Moduli Set {M-1, M}

M. I. Daabo and K. A. Gbolagade

This paper presents a new Residue Number System (RNS) overflow detection for the moduli set-{M-1, M}, where M is the system dynamic range. The algorithms are based on the computation of residues and then comparing their values. In detecting overflow with this scheme, we proposed that If the moduli set {M-1, M} is a transformation of the moduli set {m1, m2, m3,…, mn }, then for a given decimal number, X < M, there is no overflow when x1 computed is less than or equal to x0, where x0 = M-1 and x1= M. Based on this, a multiplicative and an additive overflow detection processor requiring relatively lesser hardware with faster operations when compared with the state of the art designs, is proposed. Theoretical analysis indicates that the proposed scheme outperforms the best known similar state of the art designs in terms of delay with similar area cost.

Keywords: Residue Number System, Overflow detection, Dynamic Range

Model Ranking the Classification Accuracy of Data Mining Algorithms for Anonymized Data

M. Sridhar and B. Raveendra Babu

In recent years, data has been increased in many systems and to protect such data before its release the data should be anonymized. To ensure the privacy of data, in the literature there were many techniques proposed through the privacy preserving data mining algorithms, secure multiparty and information hiding. For anonymizing the data the available methods like randomization, k-anonymous, l-diversity and t-closeness. This paper focuses on the performance of classification accuracy and computational time of the data with and without k-anonymization to rank the best data mining algorithm. The classification accuracy is calculated using Naive Bayes’, Bagging, K-nearest Neighbor, PART, oneR, and J48.

Keywords: Classification accuracy, k-anonymous, Privacy preserving data mining, k-nearest Neighbor

Timing Constraints Support on Petri-Net Model for Healthcare System Design

Sabri Mtibaa and Moncef Tagina

The worldwide healthcare organizations are facing a number of daunting challenges forcing systems to benefit from modern technologies and telecom capabilities. Hence, systems evolution through extension of the existing information technology infrastructure becomes one of the most challenging aspects of healthcare. In this paper, we present a newly architecture for evolving healthcare systems towards a service-oriented architecture. Since healthcare process exists in temporal context, timing constraints satisfiability verification techniques are growing to enable designers to test and repair design errors. Thanks to Hierarchical Timed Predicate Petri-Net based conceptual framework, desirable properties such as deadlock free and safe as well as timing constraints satisfiability can be easily checked by designer.

Keywords: Healthcare; information technology; service-oriented architecture; Hierarchical Timed Predicate Petri-Net; conceptual framework; timing constraints satisfiability

The cooperation of requirement engineering and business process modeling to elicit requirements

A. Khosravi and N. Modiri

The main goal of producing a software application is to support requirements which the software is designed for. As a software supports stakeholders’ requirements more complete and accurate, it will become more successful and accepted. Detecting business processes and their goals are inseparable part of system development. Organizations endeavor to overcome their rivals by providing distinctive services. To achieve these goals, they identify, optimize and redesign their business processes. The goals of these processes are very important since if organizations don’t elicit their goals, they won’t be able to implement them appropriately so that their projects will be failed. On the other hand, during modeling and analyzing business processes, we can understand some undiscovered requirements which they had been neglected before process modeling. In this research we will show the cooperation of requirement engineering and business process modeling to elicit system requirements.

Keywords: RE, BPM, Requirements elicitation

Reduction of the continuous Hopfield networks architecture

K. El Moutaouakil and M. Ettaouil

The Continuous Hopfield Networks (CHN) is composed of one-layer neurons with fully connected synap-ses. To calculate the equilibrium point of the CHN, many researchers use the famous Euler-Cauchy method. The complixity of this later grows with the number of neurons. The goal of this work is to reduce the size of the CHN archi-tecture. In this context, a neuron is suppressed if his vector weights and his baias fulfilled some special properties. To treat all the neurons in turn, we propose an iterative algorithm whose the convergence is quick. Furthermore, we show that the resulting network is equivalent to the initial one in terms of equilibrium points. To compare our method to the classical one, some experimental results are introduced.

Keywords: Continuous Hopfield Networks (CHN), combinatorial problems, equilibrium point, reduction of the networks architectures

Performance analysis of target tracking via wireless sensor network

Amr Lotfy Elewa M. and Osama M. El Ghandour

One of the important application of wireless sensor network is tracking mobile targets, we adopted the elliptical scheme to solve the target tracking problem by taking into account the coverage area , the quality of monitoring , path of target, conserving power and handover.We proposed a new method to minimize power required for tracking , and we where able to save the power by an amount of 9.16% and solve the problem of handover by devising the area of tracking into what is called relay area medium and relay area focal scheme respectively. By means of this we obtained the best performance of QoM and solved handoff problem even under severe Fading conditions and saving power by elliptical scheme than the other schemes.

Keywords: QoM: quality of monitoring, WSN: wireless sensor network, ellipse, relay area focal, relay area medium, Pf: probability of false alarm, Pm: probability of target missing.

Comparative Study and Performance Optimization of Vanet Using Clustering and Mesh based Approach

Ankita Anand and Parminder Singh

Vehicular ad hoc networks (VANETs) have actually attracted a lot of attention over the last few years as being used to improve road safety.In this paper Vanet has been introduced having two scenarios, one with cluster based and the another without clustering(i.e Mesh based). In cluster based technique vehicles are restricted to move in a particular direction and in the one without clustering, vehicles are not clustered, and every vehicle remains independent and free to move independently in any direction. For this the network simulator NS2 has been used.And then NS2 is using (.tr) files from MOVE.It results in a wireless Mesh based Vanet using Traffic aggregation device,wireless mesh router,wired connectivity gateway and end user device.

Keywords: Clustering, IEEE 802.11, mesh networks, Vanet

E-recruitment is an important componet of Relational E-Human Resource Management (E-HRM)

Loay Edward Gorge and Kamaran HamaAli A. Faraj

Human recruitment in Iraq-KRG (Kurdistan Region Government) were in a traditional systems. The human recruitment system suffers continually from being a slow-paced process, in which, jobseekers and employers must do lots of in the hunt for find suitable person in a traditional way. Additionally the process is time, effort, and money consuming for the recruiter and jobseeker. Due to lots of paper work, that effect unfairness decisionmaking. This research addresses the problem statement and provides the roles of internet solution for facilitate a fully automating the whole recruitment process, and eliminating the paper work and unfairness jobseeker selection. The developed solution is ARS (Automated Recruitment System); ARS assist and improve human resource management and help both employers and jobseekers via internetworking mean that to increase the speed of recruitment and decrese the corruption decisionmaking. In addition, they become vital assistance to human discrimination to put right people in right places and highly mange the human resources in better Quality of Service (QoS). In this paper, an ARS is proposed; it supplied with some recommendation tools. The system designed and implemented using 3-tier client/server architecture model. Four recommendation methods introduced, two of them are unintelligent and they mainly depend on the user’s decision; one of them depends on listing mechanism, and other depends on the navigation mechanism. The other two recommenders are intelligent; they depend on matching between the employer’s demands and jobseeker’s qualifications (plus demands) to make the jobseeker- recommendation or recruiter-decisions.

Keywords: Recruit, Internet-recruitment, OnlinARS, cyber-recruitment, Job-finder, human resource, job-seeker, Job-center

LRC: Novel Fault Tolerant Local Re-Clustering Protocol For Wireless Sensor Network

M. Nazari Cheraghlou, S. Babaie and M. Samadi

One of the most important and efficient factors in longevity of wireless sensor networks is their energy consumption. Therefore; the less energy consumption, the more networks lifetime. On another hand, fault tolerance and avoidance of Repeating operations, which are because of faults should be re-done will reduce energy consumption and cause the longevity of these networks. In this paper, we will review and introduce the fault tolerant clustering protocols like: Leach and DSC and also fault tolerant clustering protocols like: FT – LRC together with their weak points. Furthermore, by taking the advantage of these protocols, we’ll introduce a new protocol named FT-LRC, which has the whole benefits of previous protocols and has the least Energy Consumption Level. In this regard, fault tolerant is better than other previous protocols and other main advantages of these clustering protocols, which make them distinguished among the others is that it does re-clustering phase locally. While other protocols to do this matter globally, which cause procrastination and more energy consumption and reduction of network‘s total efficiency.

Keywords: Wireless sensor network, Clustering, Fault Tolerance, Network lifetime

Intelligent Intrusion Detection In Computer Networks Using Fuzzy Systems

Amin Einipour

The Internet and computer networks are exposed to an increasing number of security threats. With new types of attacks appearing continually, developing flexible and adaptive security oriented approaches is a severe challenge. Intrusion detection is a significant focus of research in the security of computer systems and networks. The security of computer networks plays a strategic role in modern computer systems. In order to enforce high protection levels against threats, a number of software tools are currently developed. In this paper, we have focused on intrusion detection in computer networks by combination of fuzzy systems and Particle Swarm Optimization (PSO) algorithm. Fuzzy rules are desirable because of their interpretability by human experts. PSO algorithm is employed as meta-heuristic algorithm to optimize the obtained set of fuzzy rules. Results on intrusion detection dataset from KDD-Cup99 show that the proposed approach would be capable of classifying instances with high accuracy rate in addition to adequate interpretability of extracted rules.

Keywords: Intrusion Detection, Fuzzy Rule Extraction, Particle Swarm Optimization(PSO) Algorithm

Design and Implementation of Network Security using Neural Network Architecture

Inadyuti Dutt, Soumya Paul and Sudipto Chandra

Over the last few years, secured transmission of data has been a major issue in data communication. This implementation mainly concerns about the security of confidential information and data transmission through Neural Network in order to provide confidentiality, authentication, integrity and non-repudiation of the messages. First, an encryption algorithm is developed and implemented to achieve the aforesaid purpose. It is basically a program that takes any plain text as input from the user and produces a cipher text which is sent to the source node of neural network. Now the cipher text is again encrypted and fragmented in successive layers of neural network’s concept and finally all the fragments are collected and decrypted at destination node to get the cipher text which was input to the source node. Then the cipher text is decrypted to get original plain text.

Keywords: Neural Network

An adaptive power aware routing in MANETs

R. Madhanmohan and K. Selvakumar

Research on power aware routing in MANET is a challenging task today. We have focused on efficient power aware routing algorithm which is described in this paper. The routing protocol named Power aware Ad hoc on demand distance vector ( PA-AODV) is proposed for low Energy consumption. Similarly shortest path issues are also solved. This protocol enables the updating in routing table for minimizing the energy consumption. PA-AODV uses an energy threshold function for filtering out the nodes with low logical residual energy, and reducing the broadcast operations in route discovery. Simulation results shows that PA-AODV performs well compared to ad-hoc on-demand distance vector (AODV) routing protocol even after introducing energy related fields in PA-AODV.

Keywords: Power aware ad hoc on-demand multipath distance vector, residual energy, shortest path, system lifetime, traffic anticipation