Rummy game agent is trained using Reinforcement Learning algorithm. It is trained without using any high-level libraries like Tensorflow, Scikit-Learn.
This game can be played online by creating a server and connecting multiple clients to the server. The trained model is imported in the client code by importing the Q-table (using .pickle) file.
The details of the code and other aspects are described in detail in ipython notebook.
Please refer to "Abhilash Mandlekar - Assign4.ipynb".