Reinforcement Learning in Python — Approximation Methods and Applications

Led by Demis Hassabis Simulacrum

Scaling reinforcement learning with function approximation — linear models, feature engineering, neural networks, CartPole, and a stock trading project using Q-Learning.

1. Module 1

   Linear Approximation and Feature Engineering

   Led by Demis Hassabis Simulacrum

   The question

   Why tabular methods fail in large state spaces · linear models for reinforcement learning · feature engineering for RL (tile coding, radial basis functions, polynomial features) · approximation methods for prediction (semi-gradient TD) · approximatio...

   Outcome

   Demonstrates understanding and implementation of linear approximation and feature engineering.

   Sub-units

   1. ○ 1.1 Linear Approximation and Feature Engineering Open this unit → ✓ ↻
   Begin the course → Print certificate of completion
2. Module 2

   Neural Networks and CartPole

   Led by Demis Hassabis Simulacrum

   The question

   CartPole environment (balancing a pole on a moving cart) · using neural networks as function approximators · CartPole with approximation methods in code · the connection to deep Q-networks (DQN) · approximation methods exercises...

   Outcome

   Demonstrates understanding and implementation of neural networks and cartpole.

   Sub-units

   1. ○ 2.2 Neural Networks and CartPole Open this unit → ✓ ↻
   Open this module → Print certificate of completion
3. Module 3

   Stock Trading with Q-Learning

   Led by Demis Hassabis Simulacrum

   The question

   Stock trading project introduction · designing the trading environment (states, actions, rewards) · how to model Q for Q-Learning in a trading context · programme design and architecture · implementing the trading agent (data loading, environment, Q-...

   Outcome

   Demonstrates understanding and implementation of stock trading with q-learning.

   Sub-units

   1. ○ 3.3 Stock Trading with Q-Learning Open this unit → ✓ ↻
   Open this module → Print certificate of completion