AI & ML
Rao-Blackwellized e-variables
arXiv stat.TH0 views
arXiv:2512.16759v2 Announce Type: replace Abstract: We show that for any concave utility, the expected utility of an e-variable can only increase after conditioning on a sufficient statistic, echoing the classic Rao-Blackwell theorem. The simplest form of the result has a straightforward proof using Jensen's inequality, but we prove more sophisticated versions. We provide several applications of this insight, including a derivation of the log-optimal e-variable for testing the variance in a linear regression model.
Source:https://arxiv.org/abs/2512.16759PreviousMultiple testing in multi-stream sequential change detectionNextA Data-Consistent Approach to Ensemble Filtering
Related posts
What tech entrepreneurs can teach scientists about failure
0Nature News

X for iPhone and iPad just added over a dozen new app icon options
09to5Mac
iOS 27.2’s clever new Messages feature can be tailored to your needs
09to5Mac

OpenAI is retiring ChatGPT Classic app, will soon combine Chat and Work tabs in new app
09to5Mac
Cilium 1.20.0 Requires BTF Flag in Linux Kernel, Raspberry Pi OS Kernel Incompatible Without It
0dev.to
YC Labs and Olam Labs ran 50 AI agents in a Minecraft civilization experiment
0dev.to