A Classification: Using Back Propagation Neural Network Algorithm to Identify Cataract Disease

Abstract

Artificial Neural Networks are often used in the fields of pattern recognition, speech, and image recognition, where high levels of computation are needed. One method that can be used is the Back-propagation Neural Network. The method can be used to identify several diseases, one of which is a cataract. A cataract is one of the most significant diseases that can cause blindness. Cataracts consist of three levels, namely mature cataracts, immature cataracts, and hyper-mature cataracts. The testing method uses two parameter values, namely epoch with value 1000,5000,10000 and learning rate with value 0.01, 0.05, 0.1. From the test, it was found that the best parameters of the above trial results were epoch with a value of 10000 and learning rate with a value of 0.1 on 80 experimental data. In experiments conducted to identify normal, mature, immature, and hyper mature obtained an accuracy of 100%, 95%, 85%, and 90%. The percentage value of the accuracy of the test results using the BPN method is 92.5%.