eprintid: 26714
rev_number: 11
eprint_status: archive
userid: 71
dir: disk0/00/02/67/14
datestamp: 2017-07-21 01:10:17
lastmod: 2017-07-21 01:10:17
status_changed: 2017-07-21 01:10:17
type: thesis
metadata_visibility: show
creators_name: KHALDA LUQYANA, NIM. 12651041
title: PENGENALAN KARAKTER TULISAN TANGAN HURUF JAWA MENGGUNAKAN METODE JARINGAN SARAF TIRUAN BACKPROPAGATION
ispublished: pub
subjects: TB
divisions: jur_tinf
full_text_status: restricted
keywords: ANN, Backpopagation, Feature Extraction Chaincode, Java Character
note: Nurochman, S.Kom., M.Kom
abstract: Java written characters is one of the traditional written characters are traditional’s Nusantara writing. To recognize handwritten letters require a method to process the image into a pattern, as well as to classify the objects that will be recognized by the computer. Given the handwriting is different from the print paper that has different forms and vary in composition. So that Java handwritten need a method to be recognized by the computer. This research method using Chaincode Feature Extraction method of image processing which is to present a line, curve, or the borders of an area. As well as Backpropagation method which is a method of artificial neural network that can be applied to both the forecasting. To obtain a balance between the ability to recognize patterns of network and network capabilities to provide a response. The test results using Backpropagation ANN is equal to 88.5% for the training data validation, and identify 5.33% on testing the test data. Optimal architecture used 60 input nodes, 2 hidden layer with 32 nodes of hidden layer on the first layer and 32 nodes in the second layer, and 5 output node. The parameters used are MSE 0001; learning rate 0.1, momentum 0.7; and 10000 epoch. Several factors can affect the size of the level of success of the introduction of them: some characters that are similar that could cause the reading patterns of letters to each other may be equal, the choice of several methods of image processing are less precise and hidden layers and hidden layer nodes, the smaller the value of nodes hidden layer, the smaller is also the result of recognition.
date: 2017-05-30
date_type: published
institution: UIN SUNAN KALIJAGA YOGYAKARTA
department: FAKULTAS SAINS DAN TEKNOLOGI
thesis_type: skripsi
thesis_name: other
citation:   KHALDA LUQYANA, NIM. 12651041  (2017) PENGENALAN KARAKTER TULISAN TANGAN HURUF JAWA MENGGUNAKAN METODE JARINGAN SARAF TIRUAN BACKPROPAGATION.  Skripsi thesis, UIN SUNAN KALIJAGA YOGYAKARTA.   
document_url: https://digilib.uin-suka.ac.id/id/eprint/26714/2/12651041_BAB-I_IV-atau-V_DAFTAR-PUSTAKA.pdf
document_url: https://digilib.uin-suka.ac.id/id/eprint/26714/1/12651041_BAB-II_sampai_SEBELUM-BAB-TERAKHIR.pdf