Volume 6 Issue 8- August 2016

 

Page Title Full Text
293-298 Data Discretization Technique Using WEKA Tool
A. Rajalakshmi, R. Vinodhini, K. Fathima Bibi
Abstract

Knowledge Discovery from Data defined as “the non-trivial process of identifying valid, novel, potentially useful, and ultimately understandable patterns in data” data pre-processing is an essential step in the process of Knowledge Discovery. The goal of pre-processing is to help improve the quality of data, consequently the mining results. Real-world data sets consist of continuous attributes. Many algorithms related to data mining require the continuous attributes need to be transformed into discrete. Discretization is a process of dividing a continuous attribute into a finite set of intervals to generate an attribute with small number of distinct values. In this paper handle continuous values of iris data set taken from UCI machine learning repository. Discretization filter applied in iris data set using WEKA Tool and also data set used in various classification algorithms namely J48, Random Forest, RepTree, Naïve Bayes, RBF network, OneR, BF Tree, and Decision Table.  The performance measures are Accuracy and Error Rate noted both before and after discretization, and it shows that discretization improves the classification accuracy in iris data set.

Keywords: Classification, Discretization, Pre-processing, WEKA Tool, Decision Tree


Full Text PDF
299-301 Generation of Hex File using Robotic Artist
Sowmya M and UshaRani J.
Abstract

Robo-Artist is an application which creates the drawing using FireBird V Robot through image processing. This is the integration of currently available robotic application with various methods of image processing. Grayscale conversion will be applied to the given drawing and Edge Detection will be performed. Execution in Matlab yields the result in Binary Matrix. To move the robot, the program is executed in the AVR studio and the result is obtained as hex file. Then hex file is burn to the FireBird V through USB cable for the movement of robot to draw an image.

Keywords: Grayscale, Edge Detection, Binary Matrix


Full Text PDF
302-305 Enhancing Network Lifetime Maximization by Efficient Routing Protocol in Wireless Sensor Network
N.Archana
Abstract

Traffic patterns in wireless sensor networks (WSNs) generally follow a many-to-one model. This paper presents how to balance power consumption and maintaining the networks lifetime for sensor network. The proposed system focus the influence of multiple mobile sink nodes on power consumption and network lifetime, and mainly focus on the selection of movable sink node number moreover the selection of parking positions, as fine as their impact on performance metrics above. For the lifetime maximization of the network is improved with the help of Ant Colony Optimization algorithm at the same time as well as Short cut tree routing algorithms are used. Without referring the router table information data could be forwarded through the shortest path. And the verification process of the data could be achieved by Short cut tree routing algorithm. By using the proposed algorithm the lifetime of the network could be increased

Keywords -- Wireless Sensor Network, Energy Conservation Scheme, Mobility, Controlled Mobility.


Full Text PDF
306-310 Bottlenose Dolphin Whistle Categorization Using Eigenwhistle Based Approach
Michael MacBride, Lucas Roth
Abstract

Dolphins produce a wide variety of whistles that vary their contour shapes in terms of frequency band and spatial length. Characterization of whistles is based on the premise that dolphins share a function-specific repertoire of whistles and each one is used in a particular behavioural context which has not been investigated adequately in the literature. Therefore, the automated categorization of whistles introduced in the paper helps marine biologist to correlate them with animal's behavioural contents. Dolphin whistles visualized in time-frequency representation are passed through a digital bandpass filter to throw off undesired and noisy information. A well-known image processing technique is adapted to process the spectrograms and extract salient features named as eigenwhistles for all training samples of different types. In the evaluation phase, testing whistles are projected on the training eigenwhistle space and based on the Euclidean metric, they are assigned to different classes of whistles. The results demonstrate the capability of eigenwhistles to correctly classify whistles with an accuracy of 90% on a total of 500 calls with 7 call types.

Keywords— Whistle categorization, Eigenwhistle, Dolphin, Whistle perception.



Full Text PDF

 

 

 

 

 

IJCSET Menu

Downloads