Machine Learning Engineer
* Top Project Award. * Used Machine Learning for condition recognition in the image dataset, three abnormal conditions were present alongside normal retinas. - Did architecture (Resnet50, Inception-v3, VGG16) and parameter selection. - Used Transfer Learning (using ImageNet and COVID-CTset as base). - Implemented Semi Supervised Learning (using only 5% of labeled images).Improved performance on a limited labeled dataset (5% labeled) using a semi-supervised learning approach: 0.9% F1 score improvement from the baseline.Retinal OCT: Retinal Optical Coherence Tomography create images by scanning the back of the eye using specified equipment. An ophthalmologist can detect different eye conditions to treat them. Detecting and treating these conditions in early stages usually means recovery and no vision loss.