CHAPTER 16 / 40
Reinforcement Learning
5 lessons · Explore, experiment and understand.
Let’s figure it out.
Ask me about this course or where to start.
- Step 71~5m3D · 1Code · 11★Most cinematic
Markov Decision Processes
A Markov Decision Process (MDP) formalizes sequential decision-making. It's the tuple : state space , action space , transition probabilitie…
Open lesson - Step 72~5m3D · 1Code · 11★Most cinematic
Bellman Equations & Value Iteration
The value function measures how good it is to be in state when following policy . The action-value function measures the value of taking act…
Open lesson - Step 73~5mCode · 11🎯Quick win
Policy Gradient Theorem
Policy gradient methods directly optimize a parameterized policy by gradient ascent on the expected return . Unlike value-based methods, the…
Open lesson - Step 74~5m3D · 1Code · 11★Most cinematic
Q-Learning and Temporal Difference
Q-Learning (Watkins 1989) learns the optimal action-value function without needing a model of the environment. After each transition it upda…
Open lesson - Step 75~5mCode · 11🎯Quick win
Actor-Critic Methods
An actor-critic algorithm trains two networks together: the actor (policy) and the critic or (value estimate). The actor improves via policy…
Open lesson