Penerapan Artificial Neural Network (ANN) untuk Memprediksi Perubahan Derajat Miopia pada Manusia

Abstract

Myopia is a vision disorder that causes the sufferers unable to see distant objects. The degree of myopia in humans can changes, both increasing and decreasing. The increasing of myopia degree is proportional to the potential of other visual disorders, such as cataracts, retinal detachment, and glaucoma. Therefore, the increasing of myopia degree needs to be watched out. Several previous studies only considered the time factor in predicting the changes of myopia degree. In fact, the changes of myopia degree also influenced by some factors that related to individual identity and behavior. This study aims to predict the changes of myopia degree in humans based on some factors that causes myopia.. This study uses data that has been scaled with the fuzzy membership function to be processed with ANN for predicting the changes of myopia degree. By ANN 6-2-3 architecture that uses 80 training data, 20 testing data, and 1 predictive data, the prediction result of the changes of myopia degree in the right eye is 1.1 dioptri, in the left eye is 1.2 dioptri and the accumulated of both is 2.3 dioptri with accuration values 87.79%, 78.47%, and 83.21%.

Prediksi Kelulusan Mahasiswa Program Studi Matematika Universitas Udayana Menggunakan Backpropagation Neural Network

“…Berdasarkan hasil yang diperoleh di setiap arsitektur BNN akan dipilih satu arsitektur dengan rataan nilai fungsi loss terkecil dengan nilai paling konstan dari seluruh arsitektur BNN (Sapitri et al, 2020) Berdasarkan tabel 4, diperoleh informasi bahwa arsitektur 5-3-1 memiliki rataan nilai Cross-Entrophy Loss terkecil dan paling stabil pada setiap percobaan. Oleh karena itu, arsitektur 5-3-1 terpilih sebagai arsitektur terbaik dan akan digunakan dalam proses prediksi data uji.…”