moreirab/dynamic-programming

Dynamic Programming mini project from the Machine Learning Engineer Nanodegree program.

Mini Project: Dynamic Programming

In this project, I wrote implementations of many classical dynamic programming algorithms.

Part 0: Explore FrozenLakeEnv

In this section, I create an instance of the FrozenLake environment.

Part 1: Iterative Policy Evaluation

In this section, I wrote my own implementation of iterative policy evaluation.

Part 2: Obtain from

In this section, I worte a function that takes the state-value function estimate as input, along with some state . It returns the row in the action-value function corresponding to the input state . That is, my function accepts as input both and , and return for all .

Part 3: Policy Improvement

In this section, I wrote my own implementation of policy improvement.

Part 4: Policy Iteration

In this section, I wrote my own implementation of policy iteration. The algorithm returns the optimal policy, along with its corresponding state-value function.

Part 5: Truncated Policy Iteration

In this section, I wrote my own implementation of truncated policy iteration.

Part 6: Value Iteration

In this section, I wrote my own implementation of value iteration.