content.tex

\maketitle

\section{introduction}
\begin{frame}
  \begin{itemize}
    \item \myhref{http://inspirehep.net/record/644725}{sPlot} is a popular ($\sim$ 650 citations) tool (at least in BaBar and LHCb)
      to assign per-event weights to candidates to disentangle signal from background
    \item useful e.g. for \myhref{http://inspirehep.net/record/1276565}{fitting}
    \item can in principle be used to train ML classifiers on data without creating a pure background and a pure signal sample
      \newline (lots of fine print skipped here)
      \item e.g. among NeuroBayes users an appreciated feature that sPlot weights are supported
  \end{itemize}
  \begin{block}{this should be an FAQ}
    \begin{itemize}
        \item In my experience it's a reoccuring question on statistics mailing lists, if some ML tool supports sPlot weights, and one then gets many answers, each covering only one tool / one method of one tool / one setting of one method of one tool
          \item \dots and to make things worse, people report from memory, which may not be up to date with the latest version of tools anymore
    \end{itemize}
  \end{block}

\end{frame}


\begin{frame}
  \frametitle{conclusion}
  \begin{exampleblock}{what I want to get done}
          work towards a somewhat comprehensive overview table of which tools support sWeights and which don't.
  \end{exampleblock}
  

  \vspace{.3\textheight}

  \IfFileExists{./QR2.png}{
  \footnotesize{slides (excl.\ cern logo) will appear on}
    
    \myhref{https://gitlab.cern.ch/pseyfert/slides-negativeweights-2017-02-28}{https://gitlab.cern.ch/pseyfert/slides-negativeweights-2017-02-28}\includegraphics[width=.2\textwidth]{./QR2.png}
}{}
\end{frame}