From 61fac3f1285d1e4bda741a4b6ff1d5f37d6681ba Mon Sep 17 00:00:00 2001
From: Jan Grewe <jan.grewe@g-node.org>
Date: Thu, 30 Oct 2014 18:10:36 +0100
Subject: [PATCH] more exercises for the reconstruction.

---
 programming/lectures/sta_stc.tex | 13 +++++++++++++
 1 file changed, 13 insertions(+)

diff --git a/programming/lectures/sta_stc.tex b/programming/lectures/sta_stc.tex
index e789e20..fee9cec 100644
--- a/programming/lectures/sta_stc.tex
+++ b/programming/lectures/sta_stc.tex
@@ -333,11 +333,24 @@ ylabel('correlation')
     method to create the PSTH) and convolve (\verb+conv+) it with the
     \textbf{STA}.
   \end{enumerate}
+\end{frame}
+
+\begin{frame}
+  \frametitle{Relating stimulus and response}
+  \framesubtitle{Reverse reconstruction using the \textbf{STA}}
   \textbf{Exercise:}
   \begin{enumerate}
   \item Reconstruct the stimulus from the P-unit data.
   \item Plot original and reconstructed stimulus into the same plot.
   \item Are they similar? In which instances are they different?
+  \item Estimate the differences in terms of the
+    \textit{mean-square-error} you have used in week 2. Normalize it
+    to the standard-deviation of the original stimulus.
+  \item Estimate the \textbf{STA} on the basis of the first half of
+    the data and reconstruct the stimulus of the second half.
+  \item Compare the reconstruction errors. 
+  \item How strongly does the error depend on the amount of data you
+    use?
   \end{enumerate}
 \end{frame}