Abstract: The well-trained image classification neural networks are vulnerable to adversarial examples. An adversarial example is a malicious input carefully crafted by adding small perturbations to ...

Abstract: In this study, a convolutional neural network (CNN)-based method for eye disease recognition is proposed, aiming to identify multiple common eye diseases through automatic analysis of fundus ...