Chapter 17: making complex decisions
ARTIFICIAL INTELLIGENCE MODERN APPROACH
Visual Summaries
.png)
Chapter 01: Introduction
ARTIFICIAL INTELLIGENCE MODERN APPROACH
Table Summary
Core Terminology
Master key academic terminology through active recall and spaced repetition concepts.
 Markov Decision Process (MDP)
A sequential decision problem for a fully observable, stochastic environment with a Markovian transition model and additive rewards.
 optimal policy (π∗)
The policy that yields the highest expected utility among all possible policies.
Value iteration
Which algorithm calculates an optimal policy by iteratively updating state utilities until they reach equilibrium?
 'contraction' mapping
An operator that, when applied to two different vectors, results in two new vectors that are closer together.
Policy evaluation and policy improvement.
What are the two alternating steps in the policy iteration algorithm?
modified policy iteration
An algorithm that approximates policy evaluation using a few steps of value iteration instead of solving linear equations exactly.