HW3 2.1 Question

65 views

Skip to first unread message

Yitong Zhou

unread,

Mar 15, 2013, 12:51:20 PM3/15/13

to 10-701-spri...@googlegroups.com

I get confused about the error 0.001, in a formula like Pr(X>\epsilon) < \delta, is this error referring to the \epsilon or the \delta, but what is the value of the other variable, I think I cannot bound the inequality with only one variable..

Thanks,

Yitong

Sam Zhang

unread,

Mar 15, 2013, 4:32:38 PM3/15/13

to 10-701-spri...@googlegroups.com

I think error is epsilon (0.001), delta is failure probability which I recall mentioned in an earlier post is set to be 5%.

Sam

Se-Joon

unread,

Mar 15, 2013, 6:04:45 PM3/15/13

to 10-701-spri...@googlegroups.com

Should we assume that \delta is for each shade of blue? Because if we have that, the probability of all 40 shades of blue to be estimated successfully is only (0.95)^40 = 0.1285. I think it makes sense to use a union bound so that we can have the probability for all 40 of them to be estimated correctly is 95%.