No the best action for achieving target one can

Info iconThis preview shows page 1. Sign up to view the full content.

View Full Document Right Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: Peeking is part of the current percept. b) A simple reflex agent is sufficient a. The agent hits unless the world state is such that a hit would cause a bust (that is, hit unless your current total + the peeked card is greater than 21) b. No. The best action (for achieving target one) can always be uniquely determined from the current percept. Thus keeping a model is doesn’t help any or change the action selected. c. Any change to the task environment that would require remembering information from past percepts (a.k.a. keeping a model) is acceptable as an answer here. c) Any pseudocode that expresses something like this is acceptable: if (CalcSum(agent) + next card) <= 21 then return “hit”/true else return “stand”/false d) Yes. The agent knows how the dealer plays (h...
View Full Document

This note was uploaded on 01/22/2014 for the course EECS 492 taught by Professor Staff during the Fall '08 term at University of Michigan.

Ask a homework question - tutors are online