Volume 7 Number 1 June 2017

1

A Combined Approach of Naive Bayes Classifier and Relevance Vector Machine for Breast Cancer Diagnosis
B. M. Gayathri, C. P. Sumathi

Abstract- Diagnosing cancer manually may have some limitations and it is difficult too. Many researches are still under process for detecting cancer accurately. Many software applications are developed to diagnose the disease and some are developed for analyzing data for effective usage. Machine learning techniques are very popular for developing medical applications. There are many techniques under machine learning for cancer detection which gives appropriate results. This article deals with diagnosing breast cancer in combination of Naïve bayes and Relevance vector machine algorithms. Wisconsin original breast cancer dataset is used for testing and training.

2

Performance Analysis of Classification Algorithms on Diabetes Dataset
K. Saravanapriya, J. Bagyamani

Abstract- The healthcare environment is generally perceived as being ‘information rich’ yet ‘knowledge poor’ (Wasan, 2006). Today in this hectic lifestyle, one of the major threats to human health is Diabetes Mellitus. Valuable knowledge can be discovered from application of data mining techniques in the Health care System particularly in Diabetes Database. There is a wealth of data available within the healthcare systems. However, there is a lack of effective analysis tools to discover hidden relationships and trends in data. Knowledge discovery and data mining have found numerous applications in business and scientific domain. This paper aims to analyse the performance of the classification techniques in diabetes data set.

3

A Survey on Gateway Selection Techniques in Wireless Ad Hoc Network
S. Usha, S. Sathish

Abstract- Wireless Network enables people to communicate and access applications and information without wires. A Wireless Ad hoc Network (WANET) is a decentralized type of wireless network. The network is ad hoc because it does not rely on a pre-existing infrastructure and access points in managed wireless networks. MANET is a type of ad hoc network that can change locations and configure itself on the fly. A Mobile Ad hoc Network (MANET) is a continuously self-configuring, infrastructure-less network of mobile devices connected wirelessly. Vehicular Ad hoc Network (VANET) is a new technology based on MANET that provides communication among vehicles which is improve traffic safety and comfort of driving and travelling. A Wireless Sensor Network (WSN) is a wireless network consisting of spatially distributed autonomous devices using sensors to monitor physical or environmental conditions. The main role of gateway node like a border node is a mobile node which has at least one neighbour belonging to a different cluster. Although Clustering has been used to reduce data propagation traffic and facilitate network management in Wireless Ad hoc Network. A gateway is a vital role for communication among Wireless Networks. Selecting optimum network and gateway is the key point in integrated networking process. The selection of gateway is main aspects in clustering and non-clustering techniques. This papers focus on various gateway selection algorithms proposed for stability and communication in Wireless Ad hoc Network.

4

Impact of Watchdog Mechanism in VANET – A Review
S. Raagavi, S. Sathish

Abstract- Vehicular Ad-Hoc Network is a type of the ad-hoc network. It is communicated from the road side units. Vehicular Ad-Hoc Network is the supper class of Mobile Ad-hoc networking in which vehicles is moving high speed on road side to exchanging information in efficient manner. Vehicular Ad-Hoc Network also provides value added services like email, audio and video sharing etc. Watchdog is a security function in journalism informs the public about society especially in circumstances where a significant portion of the public would demand changes in response. Watchdog might involve for fact checking statements of public officials. In order to provide the reliable information to the environment, the mechanism like watch dog is required to monitor the misbehaving node in the network. Lot of watch dog methods is available to monitor the environment and provide the secured transformation. This paper mainly focus the survey of watch dog methods like watchdog timer, watchdog monitoring and co-operative mechanism in Vehicular Ad-Hoc Network environment.

5

Authentication of Energy Constrained Devices in Internet of Things based Healthcare System
R. Shantha Mary Joshitta, L. Arockiam

Abstract- Internet of things is the prelude to new technological innovation. It expands with different applications such as smart home, smart city, smart grids, smart car, smart healthcare and smart retail. The application of IoT in Healthcare system brings high value for the elderly, victims of chronic disease and those who require constant supervision. But there are many security issues in such application. Moreover, the medical devices used in the IoT enabled Healthcare System are resource constrained devices. So, this paper proposes a novel mechanism for authenticating such resource constrained medical devices. New algorithm for secure authentication and key agreement of the medical devices is also presented. This mechanism is resistance against various security attacks such as eavesdropping, man-in-the-middle and Denial of Service attacks. Formal security analysis and the comparative study presented in this paper have proved that the proposed mechanism has many security features and highly secure among the already existing authentication mechanisms.

6

Online Signature Denoising using Deep Autoencoder
K. Sasirekha, R. Ravikumar, K. Thangavel

Abstract- The efficiency of an online signature authentication system depends on the quality of the online signature. Denoising of the online signature is indispensable to get a noise reduced signature. Deep learning models have been applied to a wide variety of denoising problems in recent years with great success. In this paper, deep autoencoder is proposed to use for removing Gaussian noise present in the online signature. Signal denoising can be performed with autoencoders by distorting the original signal data and add some noise to it that help in generalizing over the test set. A stacked denoising autoencoder (deep autoencoder) is a denoising autoencoder with multiple hidden layers and is trained layer by layer, by trying to minimize the reconstruction error. In this research work, experiments have been conducted on the MCYT online signature dataset. The Peak Signal to Noise Ratio (PSNR), Root Mean Square Error (RMSE), Mean Square Error (MSE), Maximum Error rate (MAXERR) and L2RAT have been computed and compared in order to evaluate the performance of the proposed method

7

Mammogram Image Segmentation using Fuzzy Markov Random Field
R. Elandhendral, K. Thangavel, K. Thangavel

Abstract- The breast cancer is the second largest cause of cancer death and the most frequently diagnosed cancer in women. Mammography is the best method for detecting breast cancer in the early stage. This work mainly focuses on the mammogram image segmentation using clustering techniques. The median filter is used for pre-processing of image and it is normally used to reduce noise in an image. Connected Component Labeling is used to remove the pectoral muscle. This paper proposed a new method Fuzzy Markov Random Field for mammogram image segmentation. It is cluster based method. Fuzzy Markov random field (FMRF) is a Markov Random Field in fuzzy space which handles fuzziness and randomness of data simultaneously. The result of the segmentation algorithms compared and analyzed using Mean Square Error (MSE) and Root Means Square Error (RMSE). It is observed that the proposed method produces better results than the existing methods. All the images used in this research have been taken from MIAS.

8

Gene Expression Data Analysis using Rough K-Means Clustering Method
P. Rajalakshmi, K. Thangavel, E. N. Sathishkumar, P. S. Raja

Abstract- Data mining is the process of finding patterns in large datasets. Cluster analysis is an important part of the data mining community. The traditional clustering algorithm is slow in convergence and sensitive to the initial value in large datasets. Data clustering plays an important role in many disciplines including data mining, machine learning, pattern recognition and other fields. Cluster analysis is a popular data analysis and data mining technology. High quality and fast clustering algorithms play a vital role for users to navigate, effectively organize and summarize the data. Data mining is the process of finding patterns in large datasets. Cluster analysis is an important part of the data mining community. The traditional clustering algorithm is slow in convergence and sensitive to the initial value in large datasets. Data clustering plays an important role in many disciplines including data mining, machine learning, pattern recognition and other fields. Cluster analysis is a popular data analysis and data mining technology. High quality and fast clustering algorithms play a vital role for users to navigate, effectively organize and summarize the data. In this research work, K-Means clustering algorithm and Rough K-means clustering algorithm have been studied. Rough K-Means algorithm deals with lower and upper boundary approximations. In this paper, four datasets have taken from National Center for Biotechnology and Information (NCBI) Gene Expression Omnibus datasets. The entire datasets contain missing values and empty space or undefined values, these all are handled by filtering methods such as Gene Variance filter, Gene Low absolute variance filter, Gene Entropy filter. Then pre-processed genes have been given as an input to K-Means and Rough K-Means clustering algorithms to cluster the similar kind of genes. Comparative analyses have performed and it observed that the Rough K means algorithm selects the highly expressed genes. The Rough K Means algorithm will generate lower and upper approximations according to the mathematical property of Rough Set Theory. Then the lower and boundary values have given as an input to the Quick Reduct algorithm to select the genes. Gene Ontology weighting methods such as Biological process, Molecular Function, and Cellular components use these selected genes for finding the Biological significance. The query gene connects the entire network.