The Epsilon-Greedy /UCB ("upper confidence bound") for MAB (Multiarmed-bandit) problem som

Posted yifan2015

tags:

篇首语:本文由小常识网(cha138.com)小编为大家整理,主要介绍了The Epsilon-Greedy /UCB ("upper confidence bound") for MAB (Multiarmed-bandit) problem som相关的知识,希望对你有一定的参考价值。

你是球队教练,现在突然要打一场比赛,手下空降三个球员,场上只能有一个出战,你不知道他们的能力,只能硬着头皮上,如何根据有限的上场时间看出哪个球员厉害,然后多让他上,从而得更多分数?

Epsilon-Greedy

supposed an k arm(slot) and set ε a little number between [0,0.1]

In short, epsilon-greedy means pick the current best option ("greedy") most of the time----(1-ε) + ε/k

but pick a random option with a small probability sometimes for other option-----(k-1)ε/k

often works as well as, or even better than, more sophisticated algorithms such as UCB

for more information about

A/B testing

Thompson sampling

see

https://towardsdatascience.com/solving-multiarmed-bandits-a-comparison-of-epsilon-greedy-and-thompson-sampling-d97167ca9a50

 

以上是关于The Epsilon-Greedy /UCB ("upper confidence bound") for MAB (Multiarmed-bandit) problem som的主要内容,如果未能解决你的问题,请参考以下文章

分数在 0 和 n 之间时蒙特卡罗树搜索的 UCB 公式

DelphiLinux版的IsBadReadPtr

UCB DS100 讲义《数据科学的原理与技巧》校对活动正式启动 | ApacheCN

Linux问题:触摸屏驱动运行出现insmod: cannot insert `ucb1x00-ts.ko': File exists (-1): File exists

NRF2-KEAP1信号通路上,期待肿瘤抑制因子GULP1乘风破浪

MSP430中断的一个细节问题