Timeline for Variance of Multivariate KDE
Current License: CC BY-SA 4.0
8 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| S Mar 23, 2020 at 23:00 | history | bounty ended | CommunityBot | ||
| S Mar 23, 2020 at 23:00 | history | notice removed | CommunityBot | ||
| Mar 16, 2020 at 0:00 | history | tweeted | twitter.com/StackStats/status/1239340825739395072 | ||
| Mar 15, 2020 at 23:01 | comment | added | Syd Amerikaner | well, I tried to adapt the steps from the univariate case. I don't think $\nabla^2 f$ is bounded (math.stackexchange.com/questions/482934/…). Regarding $o$ and $O$. AFAIK there are two ways to state the taylor approximation: 1. $f(x) = f(a) + (x - a)'\nabla f(x) + \color{red}{0.5(x - a)'\nabla^2f(x)(x - a)} + \color{red}o(\Vert x - a\Vert^{\color{red}2})$, 2. $f(x) = f(a) + (x - a)'\nabla f(x) + \color{red}O(\Vert x - a\Vert^{\color{red} 2})$ (se math.stackexchange.com/a/3413546/273275) | |
| Mar 15, 2020 at 21:42 | comment | added | David Veitch | In noticed in the second order Taylor Approximation for the univariate case (bookdown.org/egarpor/NP-UC3M/kde-i-asymp.html#thm:kdebiasvar) it is $o$ not $O$. Are you sure it should be $O$? Also by assumption 51. of (bookdown.org/egarpor/NP-UC3M/kde-ii-asymp.html) I think $\nabla^2$ is bounded? | |
| S Mar 15, 2020 at 21:29 | history | bounty started | Syd Amerikaner | ||
| S Mar 15, 2020 at 21:29 | history | notice added | Syd Amerikaner | Draw attention | |
| Mar 13, 2020 at 19:32 | history | asked | Syd Amerikaner | CC BY-SA 4.0 |