Enhance Eye Disease Detection using Learnable Probabilistic Discrete Latents in Machine Learning Architectures

This is the official code repository for the project "Enhance Eye Disease Detection using Learnable Probabilistic Discrete Latents in Machine Learning Architectures". You can find the pre-print of the paper here.

This work is adapted from GFNDropout.

Steps to run

We recommend using a virtual environment to run this project. Also, a system with a GPU is recommended.

1. Install all the dependencies

pip install -r requirements.txt

2. Run the experiment script

python rfmid_gfn.py

You can make changes to the experiment settings by changing this master script. New datasets can be added by adding corresponding PyTorch classes in the data folder and exporting them for use. Along with this, Options for running the project can be edited in either the main class, located at utils/options, or can be set while the object is being initialized.