Improving Efficiency in Diabetes Detection Using Neural Network Technique
Abstract: Diabetes mellitus is an interminable disease that forces excessively high human, social and financial expenses for a nation. Additionally, minimizing its commonness rate and in addition its excessive and risky confusions requires viable administration. This paper is an effort to plan and execute a descriptive data mining approach and to devise association standards to predict diabetes behavior in arrangement with particular life style parameters, including physical activity and emotional states, especially in elderly diabetics using Neural Network. In our work network classifier has been used with different test parameters and it was found that it is effective in diagnosis of Diabetes mellitus when the person provide the required attributes value. The dataset was taken from diabetes database Indian PIMA from the UCI Machine Learning Database. The dataset is comprise of eight features which are vital in diagnosis for diabetes detection. The System is model on multilayer neural network trained with back-propagation and simulated on feed-forward neural network.

Authors: Vivek Vaidya, Dr. L. K. Vishwamitra

File Name: Vivek_90200-20-101.pdf
 Download Abstract | Download Full Paper

Bacterial Foraging Optimized Clustering Probability Approach in Heterogeneous Environment
Abstract: Sensor nodes in WSN are powered with a battery. Sensor nodes consume the battery power mainly in the tasks like data transmission, data reception and sensing. Sometimes it is impractical to replace a battery in WSN because humans can’t reach. Therefore once energy or computational resources are consumed, immediate recovery of these resources is a complex task so it is necessary to make use of battery power efficiently to increase the lifetime of the sensor nodes that will also increase the lifetime of the whole network. To make WSN energy efficient and to increase the lifetime of the network we design a Bacterial Foraging optimized clustering probability so as to find a method which increases the lifetime and reduces the energy consumption of the network. The execution and demonstration of this paper is performed with the help of MATLAB 2014a. The performance comparison metrics are; network lifetime, network throughput and number of alive nodes.

Authors: Vidya Badwaik, Prof. Rupesh Dubey

File Name: Vidya_81000-20-102.pdf
 Download Abstract | Download Full Paper

Performance Evaluation of Data Mining Algorithms used in Medical Decision Support Systems
Abstract: The main focus of the research study was to identify the most suitable & appropriate data mining algorithm used, implemented in current Medical Decision Support Systems, and also to analyze, evaluate the performance and interpret the results obtained by applying on some medical datasets. According to the previous studies and secondary data analysis three algorithms were found to be appropriate these were C4.5 (J48), Multilayer Perceptron and Naïve Bayes. The C4.5 algorithm for building decision trees is implemented in Weka as a classifier called J48. The different datasets were chosen for assessment, these five UCI databases were Kidney disease, Thoracic Surgery, Autistic Spectrum Disorder Screening Children, Statlog (Heart) and Immunotherapy Dataset. For the analysis various performance metrics or measures were utilized which includes percent of correct classifications instances, True/False Positive rates, absolute mean error, relative root squared error & other set of errors, AUC, Precision, Recall, F-measure, true positive, false positive, false negative and true negative values.

Authors: Neha Sharma, Komal Paliwal

File Name: Neha_90200-20-103.pdf
 Download Abstract | Download Full Paper

© 2016 IJDACR Journal. All rights reserved.