Institusion
Institut Teknologi Perusahaan Listrik Negara
Author
Alifa, Reysha
Luqman, Luqman
Kusuma, Dine Tiara
Subject
Teknik Informatika
Datestamp
2023-06-26 06:23:13
Abstract :
Artificial neural networks have spread everywhere and are used in applications,
one of which is the application of character shapes from hand applications. Various
patterns on the character letters have various handwriting models made by the author.
The problem that arises in the process of hand recognition is how an introduction can
recognize various letters with size, thickness, shape, and slope between one author and
another. One of the algorithms on an artificial neural network is backpropagation which
has undergone several significant changes made by a neural network that is continuously
leveled up to a certain level to be able to generalize. The test results of the block letter
handwriting character recognition system from 10 tests only 1 letter that was not legible,
namely the letter D. The test resulted in the lowest error of 0.066672. The results of the
test are 90% this means that the level of accuracy between the test data and produces
90% this means that the level of accuracy of the test accuracy between the test data and
the training data yields 90%. The implementation of the Backpropagation Neural
Network method with small epochs in carrying out something in a pattern can cause
errors that occur during training. By applying the Backpropagation Neural Network
method, you will get a higher success rate for training than new data