Neural Network Value Function Approximation

This pre-release contains the code for value function approximation with neural networks. Unfortunately, the current implementation is excruciatingly slow and you should not consider this functional.

Update: 11/18/2020
Tag also being used as easy reference before major rewrites.

Linear Function Approximaters

At this point, q-learn.py contains the code for approximating the value function with a parameterized linear function. Unfortunately, this function fails to converge appropriately and was quickly abandoned in favor of neural networks.

This release may not be stable but is mostly just so I can go back to this code since it has now been overwritten.