BerriJ/PHD-Presentation
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Improve crps theory slide

Browse Source
This commit is contained in:
Jonathan Berrisch
2025-06-27 09:44:14 +02:00
parent 99e7b2fd7b
commit 13c238779e
2 changed files with 5 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
index.html
View File

@@ -26298,12 +26298,12 @@ w_{t,k}^{\text{Naive}} = \frac{1}{K}\label{eq:naive_combination}
<p>EWA satisfies optimal selection convergence <span class="math inline">\(\eqref{eq_optp_select}\)</span> in a deterministic setting if:</p> <p>EWA satisfies optimal selection convergence <span class="math inline">\(\eqref{eq_optp_select}\)</span> in a deterministic setting if:</p>
<p><i class="fa fa-fw fa-triangle-exclamation" style="color:var(--col_amber_9);"></i> Loss <span class="math inline">\(\ell\)</span> is exp-concave</p> <p><i class="fa fa-fw fa-triangle-exclamation" style="color:var(--col_amber_9);"></i> Loss <span class="math inline">\(\ell\)</span> is exp-concave</p>
<p><i class="fa fa-fw fa-triangle-exclamation" style="color:var(--col_amber_9);"></i> Learning-rate <span class="math inline">\(\eta\)</span> is chosen correctly</p> <p><i class="fa fa-fw fa-triangle-exclamation" style="color:var(--col_amber_9);"></i> Learning-rate <span class="math inline">\(\eta\)</span> is chosen correctly</p>
<p>Those results can be converted to <em>any</em> stochastic setting <span class="citation" data-cites="wintenberger2017optimal">Wintenberger (<a href="#/references" role="doc-biblioref" onclick>2017</a>)</span>.</p>
<p>Optimal convex aggregation convergence <span class="math inline">\(\eqref{eq_optp_conv}\)</span> can be satisfied by applying the kernel-trick:</p> <p>Optimal convex aggregation convergence <span class="math inline">\(\eqref{eq_optp_conv}\)</span> can be satisfied by applying the kernel-trick:</p>
<p><span class="math display">\[\begin{align} <p><span class="math display">\[\begin{align}
\ell^{\nabla}(x,y) = \ell&#39;(\widetilde{X},y) x \ell^{\nabla}(x,y) = \ell&#39;(\widetilde{X},y) x
\end{align}\]</span></p> \end{align}\]</span></p>
<p><span class="math inline">\(\ell&#39;\)</span> is the subgradient of <span class="math inline">\(\ell\)</span> at forecast combination <span class="math inline">\(\widetilde{X}\)</span>.</p> <p><span class="math inline">\(\ell&#39;\)</span> is the subgradient of <span class="math inline">\(\ell\)</span> at forecast combination <span class="math inline">\(\widetilde{X}\)</span></p>
<p>Those results can be converted to <em>any</em> stochastic setting <span class="citation" data-cites="wintenberger2017optimal">Wintenberger (<a href="#/references" role="doc-biblioref" onclick>2017</a>)</span></p>
</div><div class="column" style="width:4%;"> </div><div class="column" style="width:4%;">
</div><div class="column" style="width:48%;"> </div><div class="column" style="width:48%;">

6
index.qmd
View File

@@ -1215,15 +1215,15 @@ EWA satisfies optimal selection convergence \eqref{eq_optp_select} in a determin
<i class="fa fa-fw fa-triangle-exclamation" style="color:var(--col_amber_9);"></i> Learning-rate $\eta$ is chosen correctly <i class="fa fa-fw fa-triangle-exclamation" style="color:var(--col_amber_9);"></i> Learning-rate $\eta$ is chosen correctly
Those results can be converted to *any* stochastic setting @wintenberger2017optimal.
Optimal convex aggregation convergence \eqref{eq_optp_conv} can be satisfied by applying the kernel-trick: Optimal convex aggregation convergence \eqref{eq_optp_conv} can be satisfied by applying the kernel-trick:
\begin{align} \begin{align}
\ell^{\nabla}(x,y) = \ell'(\widetilde{X},y) x \ell^{\nabla}(x,y) = \ell'(\widetilde{X},y) x
\end{align} \end{align}
$\ell'$ is the subgradient of $\ell$ at forecast combination $\widetilde{X}$. $\ell'$ is the subgradient of $\ell$ at forecast combination $\widetilde{X}$
Those results can be converted to *any* stochastic setting @wintenberger2017optimal
::: :::