Marek Gluza<p><a href="https://arxiv.org/abs/2101.12223" rel="nofollow noopener" target="_blank"><span class="invisible">https://</span><span class="">arxiv.org/abs/2101.12223</span><span class="invisible"></span></a></p><p>I'm city hopping by <a href="https://mathstodon.xyz/tags/train" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>train</span></a> from Warsaw to <a href="https://mathstodon.xyz/tags/Benasque" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>Benasque</span></a> and yesterday I met Oli Reardon-Smith in Kraków, here a paper with Hakop, Kamil and Stephen<br><a href="https://arxiv.org/abs/2101.12223" rel="nofollow noopener" target="_blank"><span class="invisible">https://</span><span class="">arxiv.org/abs/2101.12223</span><span class="invisible"></span></a><br>I learned about an innovation in that they use part of samples to estimate certain parameters needed for a large deviation bound. It's like <a href="https://mathstodon.xyz/tags/Hoefding" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>Hoefding</span></a>'s inequality on wheels.</p><p>I know of the <a href="https://mathstodon.xyz/tags/BCa" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>BCa</span></a> (biased corrected accelerated) <a href="https://mathstodon.xyz/tags/bootstrap" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>bootstrap</span></a> method for assigning confidence intervals to <a href="https://mathstodon.xyz/tags/statistical" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>statistical</span></a> estimations. Here a round of resampling is used to estimate two parameters to correct bias and accelerate sample efficiency in pinning down the accurate confidence interval.</p><p>What are other methods where you sample to get information about what to sample subsequently?</p>