Skip to main content
added 26 characters in body
Source Link
John Doucette
  • 9.4k
  • 1
  • 18
  • 52

Thinking about this more, the answer is in fact yes, but not for the application you mention.

You cancannot use Alphaalpha-Betabeta pruning any timeto learn a model to predict customer outcomes, because it is only useful for domains where you are performingconcerned about an heuristic-guided searchadversary.

Many, many In finding a customer model, different kindsthere is no reason to worry about someone coming in and forcing you to make bad decisions about the optimization of problems can be represented as heuristic-guided search problemsthe model. This representationConsequentially, there is not always a good choiceno reason to use minimax search, but sometimes it works very welland thus, to use alpha-beta pruning. Natural examples are:

  • Automated Theorem Proving
  • Planning and Scheduling
  • Pathfinding and navigation
  • Some kinds of optimization problem

Less natural examplesThere are applications other than (video) games where you could use these techniques though. For example, there are security games. In these "games" we want to use AI to find a strategy to protect an airport. It is reasonable to try and design our model under the assumption that someone else wants to break it. You could use Alpha-Beta pruning here (although in practice, more sophisticated algorithms are still sometimes seen today:used).

You can use Alpha-Beta pruning any time you are performing heuristic-guided search.

Many, many, different kinds of problems can be represented as heuristic-guided search problems. This representation is not always a good choice, but sometimes it works very well. Natural examples are:

  • Automated Theorem Proving
  • Planning and Scheduling
  • Pathfinding and navigation
  • Some kinds of optimization problem

Less natural examples that are still sometimes seen today:

Thinking about this more, the answer is in fact yes, but not for the application you mention.

You cannot use alpha-beta pruning to learn a model to predict customer outcomes, because it is only useful for domains where you are concerned about an adversary. In finding a customer model, there is no reason to worry about someone coming in and forcing you to make bad decisions about the optimization of the model. Consequentially, there is no reason to use minimax search, and thus, to use alpha-beta pruning.

There are applications other than (video) games where you could use these techniques though. For example, there are security games. In these "games" we want to use AI to find a strategy to protect an airport. It is reasonable to try and design our model under the assumption that someone else wants to break it. You could use Alpha-Beta pruning here (although in practice, more sophisticated algorithms are used).

Source Link
John Doucette
  • 9.4k
  • 1
  • 18
  • 52

You can use Alpha-Beta pruning any time you are performing heuristic-guided search.

Many, many, different kinds of problems can be represented as heuristic-guided search problems. This representation is not always a good choice, but sometimes it works very well. Natural examples are:

  • Automated Theorem Proving
  • Planning and Scheduling
  • Pathfinding and navigation
  • Some kinds of optimization problem

Less natural examples that are still sometimes seen today: