top of page
Chapter 21: reinforcement learning
ARTIFICIAL INTELLIGENCE MODERN APPROACH
Visual Summaries
.png)
Table Summary
Core Terminology
Master key academic terminology through active recall and spaced repetition concepts.
optimal policy
A policy that maximizes the expected total reward over time.
A reflex agent
What type of agent learns a direct mapping from states to actions?
Policy
Passive learning differs from active learning in that the agent's _____ is fixed.
reward-to-go
The observed total reward obtained from a specific state until the end of a trial.
Bayesian reinforcement learning.
What reinforcement learning approach assumes a prior probability for each possible environment model?
Prioritized sweeping
What heuristic in ADP ranks adjustments to prioritize states whose successors have just undergone large utility changes?
Chapter Quiz
According to the source, which capability is uniquely required for a computer to pass the 'Total Turing Test' compared to the standard Turing Test?
What is a primary obstacle to the 'laws of thought' (logicist) approach to artificial intelligence?
Which field is described as combining probability theory with utility theory to provide a framework for decisions made under uncertainty?
The 'General Problem Solver' (GPS) was the first program designed to embody which specific approach to AI?
Why was the DENDRAL program significant in the history of AI development?
Answers: D, C, A, D, A
bottom of page