Stock market price simulation and prediction based on Generative Adversarial Network in Deep Learning and Reinforcement Learning

Deep learning for lung pre-malignant lesions

Segmentation of epithelium in H&E stained lung issue was performed on more than 40 Whole Slide Images(WSI) using masks generated from pathologist annotations on the WSI.

A deep learning algorithm, UNet, that has a pertained VGG19 contracting path and an expansive path, was used to train the WSI at a patch level along with the binary mask of the annotation.

Using a single UNet with one hot encoder mask fro multi class segmentation.