High Availability Infra

Using machine learning to generate circuit design given properties requirement

Train machine learning models to predict glaucoma surgery outcomes

Implement U-Net to do semantic segmentation and different types of CNN architecture like ResNet, CapsNet to assist with the prediction

Use transfer learning, hyperparameter tuning, cross validation to check and improve model performance

Establish a modular, scalable training pipeline including data labeling, data preprocessing, model training and model evaluation