A Code-Driven Introduction to Reinforcement Learning
summary
Builds a 5x5 GridWorld MDP in Python and provides a Jupyter notebook (pure Python) that runs in Colab with no extra libraries, encoding start at top-left and goal at bottom-right and using reward = -1 per step. Implements random trajectory sampling, a cross‑entropy baseline and a Monte Carlo estimator, runs ~1,000 episodes to estimate state‑action values, prints ASCII visualizations and extracts a greedy policy; examples include taxi routing and HTML inbox automation.
AI-written summary · about
chapters (7)
- 0:00Intro
- 2:53Agenda
- 3:08What is Reinforcement Learning (RL)?
- 8:52Coding the MDP
- 10:35Coding the RL solution
- 29:50Next steps
- 31:53Outro
related talks
28mDeep Reinforcement Learning: theory, intuition, code
26mHacking Reinforcement Learning
56mAn Introduction to Reinforcement Learning - Snake Your Way Out of a Paper Bag
01h 17mTraining intelligent game agents using deep reinforcement learning
01h 31mWorkshop on developing Reinforcement Learning
01h 01mReinforcement Learning: Pac-Man
44mReinforcement learning: A gentle introduction
01h 01mA Tutorial on Reinforcement Learning I