KHOTIMAH, 100941002 (2011) PENERAPAN METODE JARINGAN SYARAF TIRUAN (ARTIFICIAL NEURAL NETWORK) PADA FAKTOR YANG MEMPENGARUHI TERJADINYA TBC PARU. Thesis thesis, UNIVERSITAS AIRLANGGA.

Preview	Text (ABSTRAK) gdlhub-gdl-s2-2012-khotimah-18223-tkm2311.pdf Download (309kB) | Preview
	Text (FULLTEXT) gdlhub-gdl-s2-2012-khotimah-18223-tkm2311.pdf Restricted to Registered users only Download (1MB) | Request a copy

Abstract

One method of classification is often used is discriminant analysis and logistic regression. In the discriminant analysis are the assumptions that must be met by the data, and not all data can meet the assumption. While logistic regression requires complete data, sensitive to outliers and the results of the classification be the probability that the outcome less than the maximum. Because we need analytical methods with good classification accuracy. Artificial neural network method is a method that can be used for classification with good accuracy. The purpose of this research was to determine architecture neural network that the maximum and know the size of the ability of artificial neural networks in classifying data. Based on the results of research using neural networks with backpropagation method of training data obtained maximum results. The architecture neural network that is obtained, consisting of 1 input layer with 7 nodes, 1 hidden layer with 10 hidden nodes, and 1 output layer with 1 node. That architecture resulted in classification accuracy with MSE values of 1.14732 e-005 is reached at epoch 11 with learning rate parameters used in 0:25, and the target error 0.0001. These results indicate that no data object is wrong in classifying 100% of objects can be classified correctly. Tests carried out with the new data network that produces MSE value of 1.4287 e-008 achieved at 6 epochs and 100% of data can be recognized with appropriate data classification target.

Actions (login required)

View Item