Project: Extending the Graphical User Interface for OpenAI Gym ()
Description: This project aims to extend VisualRLComposer, a GUI for OpenAI Gym which involves an interface allowing users to design reinforcement learning experimentations with predefined modules that are compatible with the OpenAI Gym library by a visual drag & drop system.
Overall Project Goals:
- Adding a tab where users can design custom CNN/ANN policies visually
- Adding support for parallel multi-environment experimentation
- Extend the training part of platform by allowing users to keep track of the training, step by step, via Tensorboard
- Integrating the GUI with an open-source platform towards designing translatable intelligent closed-loop neuromodulation systems called NeuroWeaver
Contributor: Mehul Sinha
Organization: Department of Biomedical Informatics (BMI), Emory University
Mentors:
- Özgür Kara
- Babak Mahmoudi
- Parisa Sarikhani
- Work done on first 3 project goals and other improvements have been committed to branch
, which has then been merged to main branch with
.
- Work done on integrating NeuroWeaver with VisualRLComposer is done on
branch.
From the initial proposal, through the community bonding period and in these final few days, it's honestly been a pleasure working on this project. I’ve had a wonderful time during these 4-5 months and have learned plenty of things. Special thanks to my mentor, Özgür Kara for his constant support throughout GSoC '22.
Thanks Mehul, great work!