Talk:Probability matching

Statistics Stub‑class Low‑importance

	This article is within the scope of WikiProject Statistics, a collaborative effort to improve the coverage of statistics on Wikipedia. If you would like to participate, please visit the project page, where you can join the discussion and see a list of open tasks.StatisticsWikipedia:WikiProject StatisticsTemplate:WikiProject StatisticsStatistics
Stub	This article has been rated as Stub-class on Wikipedia's content assessment scale.
Low	This article has been rated as Low-importance on the importance scale.

Robotics Stub‑class Low‑importance

	This article is within the scope of WikiProject Robotics, a collaborative effort to improve the coverage of Robotics on Wikipedia. If you would like to participate, please visit the project page, where you can join the discussion and see a list of open tasks.RoboticsWikipedia:WikiProject RoboticsTemplate:WikiProject RoboticsRobotics
Stub	This article has been rated as Stub-class on Wikipedia's content assessment scale.
Low	This article has been rated as Low-importance on the project's importance scale.

There are

There are cases where probability matching isn't suboptimal. For example, when deciding to select once of several caches of a resource (like food) and a large number of competitors, it is often better to match than maximize. For the person who de-stubs this article, I think that caveat would be useful to add.128.32.245.203 (talk) 04:05, 12 August 2010 (UTC)[reply]

Optimal? Suboptimal?

The word suboptimal seems that it does not belong here without an appropriate citation. It is my belief that there are many cases, especially in the multi-armed bandits context, where probability matching provides a solution that is at least asymptotically optimal, if not finite time optimal. See, at a minimum, Scott (2010) in Applied Stochastic Models for Business and Industry. DOI: 10.1002/asmb:

"This article describes a heuristic for managing multi-armed bandits called randomized probability matching, which randomly allocates observations to arms according the Bayesian posterior probability that each arm is optimal."

Refrozen (talk) 04:27, 15 November 2014 (UTC)[reply]

I agree with these comments. More generally, the connection between probability matching and Thompson sampling could be improved. See my related comment on the talk page of Thompson sampling. Thanks! Koedinger (talk) 13:23, 28 October 2016 (UTC)[reply]