Published 2012
“…The objectives of this paper are to develop useful
algorithms and to build a new classification model for the case study.The first approach of neural network includes training on
algorithms of Resilient Backpropagation,Scaled Conjugate Gradient Backpropagation and Levenberg-Marquardt.The outputs are used in comparison with Bayesian Neural Network for efficiency
comparison.The results showed that Resilient Backpropagation have the consistency in all measurement but performs less in accuracy.In second approach,the Bayesian Classifier_01 outperforms the Resilient Backpropagation by successfully increasing the overall prediction accuracy by 16.0%.The Bayesian Classifier_02 is built to improve the accuracy by adding new features of chemical properties as
selected by the Information Gain Ratio method,and increasing the length of the
window sequence to 201.The result shows that the built model
successfully increases the accuracy by 96.0%.In comparison,the Bayesian model outperforms Tikole and Sankararamakrishnan (2008) by increasing the sensitivity by 10% and specificity by 26%. …”
Get full text
Get full text
Get full text
Article