A Simple Loop For Bandit Problem

Question

Zahra kamkar 2014-5-13

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/129421-a-simple-loop-for-bandit-problem

回答： Manikanta Aditya 2022-6-20

Hello. I wanna write a code for Bandit problem. My problem: We have 5 machines. We wanna play each of these machines 1000 times (we call each 1000 plays a TASK). In each time, the machine gives us a reward (randomly). We wanna check that the mean reward of which one of these TASKs is more than others (maximum). I wrote this code but it doesn't work well. Why?

    for j=1:5
h=0;
for i=1:1000
    rew=randn(1);
    sum=rew+h;

end

mean(j)=sum./1000; [value,index]=max(mean(j))

end end

Thanks in advance

0 个评论
显示 -2更早的评论隐藏 -2更早的评论

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Manikanta Aditya 2022-6-20

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/129421-a-simple-loop-for-bandit-problem#answer_989320

Hi Zahra,

The calculation of mean is to be done in the inner loop, where as I see it is being done in the outer loop in your code. This is the reason for the errors. Please refer to the code below :

mean = [];

for i = 1 : 5

m = 0;

for j = 1 : 1000

s = 0;

r = randn(1);

s = s + r;

m = s / 1000;

end

mean(i) = m;

end

value = max(mean);