For practical uses we can use GPI, but the fact it actually converges is an open problem according to the textbook. Have there been any progress on this, or in practice is Monte Carlo a sound method?
by Justin
1
Vote
Alex
Still can't access Mondays slides through the website. 404 Error
How many examples per state / state action pair are usually required for Monte Carlo to converge (approximately) to the actual state value or action value. And how can you know when you are close?
0
Vote
can you reexplain why you wouldnt have q values for some states
0
Vote
What if I don't have any friends then I can model it as a single state of "don't hang out with friends" and a single action of "cry" 😛
by Your name (optional)
0
Vote
can you post the solutions - the green screen is tough to take a photo of