multi-exit block

poster.tex

@@ -212,9 +212,26 @@ Object categorization methods are trained to recognize \textbf{1000s of classes}
 \end{block}
 
 \begin{block}{\Large Example: Multi-output Distillation}
-  \begin{center}
-  \includegraphics{multi-output/architecture.pdf}
-  \end{center}
+  \begin{columns}
+    \begin{column}{0.68\textwidth}
+      \includegraphics{multi-output/architecture.pdf}
+    \end{column}
+    \begin{column}{0.28\textwidth}
+      \textbf{Multi-exit architectures}
+      \begin{itemize}
+      \item a crude initial prediction that is gradually refined
+      \item can be stopped anytime to provide a valid prediction
+      \end{itemize}
+      \textbf{Standard training}
+      \begin{itemize}
+      \item sum of exit-wise losses
+      \end{itemize}
+      \textbf{Proposed training}
+      \begin{itemize}
+      \item distillation from later (more accurate) to earlier exits
+      \end{itemize}
+    \end{column}
+  \end{columns}
 \end{block}
 
 \end{column}