" Diabetic retinopathy (DR) is a major cause of vision impairment in diabetic patients. In recent years, 28.5% of the adult U.S. population with diabetes contracted the complication, which placed 4.2 million individuals at risk of becoming forever blind [1]. It is thus crucial to be able to detect DR during onset to prevent such potentially extreme consequences of diabetes. We propose a solution to this archetypal machine learning classification problem with an Inception-based neural network and a complementary network visualization with salience maps for transparency and insight into our model’s learning. "
continued in literature review