Easy21
Assignment from David Silver's Reinforcement Learning course. Coded for clarity, not efficiency.
Requires Torch7 with the Moses package.
Run monte-carlo.lua first to generate Q* and the plot of V (below), then sarsa-lambda.lua and lin-fun-approx.lua to generate their plots.
Includes an additional method without value functions - policy-gradient.lua - that uses a simple neural network.
