You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Check "GDICE_TechReport.pdf" available in this repository for more details about the reimplemented G-DICE algorithm and the validation experiments performed
This is a re-implementation of the G-DICE algorithm based on the paper: "The Cross-Entropy Method for Policy Search in Decentralized POMDPs" by FA Oliehoek et al.
The algorithm is run on the dectiger problem from the "problems/" folder
Compile using:
sudo python3 setup.py develop
Then execute :
python3 gdice.py
The policy graphs are viewable with the installation of "xdot". Once the tool is installed, execute "xdot gdice_best_policy_agent{x}.dot" on the ".dot" files available in the "src/" folder.