6

I'm a bit confused with extensive number of different Monte Carlo methods such as:

I won't ask for the exact differences, but why are all of them called Monte Carlo? What do they all have in common? Can they all be used for AI? E.g. which one can be used for gaming (like Go) or image recognition (resampling)?

Mithical
  • 2,885
  • 5
  • 27
  • 39
kenorb
  • 10,423
  • 3
  • 43
  • 91
  • 2
    I found [this paper](http://www.cameronius.com/cv/mcts-survey-master.pdf) which looks pretty comprehensive, and they touch on Go on p.27. However, that part is outdated since [AlphaGo](https://gogameguru.com/i/2016/03/deepmind-mastering-go.pdf). You might want to look at [Matthew Lai's Giraffe Chess](https://arxiv.org/pdf/1509.01549v2.pdf) paper. Similar method, but a less complex game. – DukeZhou Aug 24 '16 at 01:42

1 Answers1

7

They are all called Monte Carlo because all of them are a different version of the canonical Monte Carlo algorithm.

The canonical version of Monte Carlo algorithm is a stochastic algorithm to determine an action based in a tree representation. The differences among all these version are their exploration and exploitation mechanisms, and it is necessary to analyse each of them to define which one fits in your case.

Mithical
  • 2,885
  • 5
  • 27
  • 39
Pedro Narloch
  • 86
  • 4