Reinforcement Learning in Python — Dynamic Programming

Led by Edsger Dijkstra Simulacrum

Solving MDPs with complete knowledge — iterative policy evaluation, policy improvement, policy iteration and value iteration, implemented in Gridworld.

1. Module 1

   Iterative Policy Evaluation and Policy Iteration

   Led by Edsger Dijkstra Simulacrum

   The question

   Dynamic programming section introduction · iterative policy evaluation (repeated Bellman updates until convergence) · designing the RL programme architecture · implementing Gridworld in code · iterative policy evaluation in code · windy Gridworld var...

   Outcome

   Demonstrates understanding and implementation of iterative policy evaluation and policy iteration.

   Sub-units

   1. ○ 1.1 Iterative Policy Evaluation and Policy Iteration Open this unit → ✓ ↻
   Begin the course → Print certificate of completion
2. Module 2

   Value Iteration

   Led by Edsger Dijkstra Simulacrum

   The question

   Value iteration (combining policy evaluation and improvement into a single update) · value iteration in code · comparison of policy iteration and value iteration (convergence speed, computational cost) · dynamic programming summary · when DP is appli...

   Outcome

   Demonstrates understanding and implementation of value iteration.

   Sub-units

   1. ○ 2.2 Value Iteration Open this unit → ✓ ↻
   Open this module → Print certificate of completion