| 123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231 |
- %!TEX root = Sommerakademie-2015-Forschung.tex
- \section{Einleitung}
- \subsection{Was ist On-Line Recognition?}
- \begin{frame}{Demo}
- \begin{figure}[h]
- \centering
- \includegraphics*[width=0.7\linewidth, keepaspectratio]{images/Classification.png}
- \end{figure}
- \href{http://write-math.com}{write-math.com}
- \end{frame}
- \begin{frame}{Was ist On-Line Recognition?}
- \medskip
- \begin{columns}[t,onlytextwidth]
- \begin{column}{.5\textwidth}
- {\Large Off-line Recognition}
- \begin{figure}[h]
- \centering
- \includegraphics*[width=0.7\linewidth, keepaspectratio]{images/A-pixel.png}
- \end{figure}
- \end{column}
- \begin{column}{.5\textwidth}
- {\Large On-line Recognition}
- \begin{figure}[h]
- \centering
- \includegraphics*[width=0.7\linewidth, keepaspectratio]{images/A-vektor.png}
- \end{figure}
- \end{column}
- \end{columns}
- \end{frame}
- \begin{frame}{Was wollen wir?}
- \[f(\text{Merkmale}) = \begin{pmatrix}0.7\\ 0.1\\ 0.2\end{pmatrix} = \begin{pmatrix} \mathbb{P}(\gamma)\\ \mathbb{P}(\text{ö})\\ \mathbb{P}(\heartsuit) \end{pmatrix}\]
- \medskip
- \visible<2->{
- \begin{center}
- {\Large Gesucht: Funktion $f$}\\
- (und Merkmalsextraktion)
- }
- \end{center}
- \end{frame}
- \begin{frame}{Merkmalsextraktion}
- \begin{figure}[h]
- \centering
- \includegraphics*[width=0.7\linewidth, keepaspectratio]{images/A-vektor-merkmalsbildung.png}
- \end{figure}
- Merkmalsvektor fester Länge ist praktisch
- \end{frame}
- \section{Funktionen}
- \subsection{Funktionen}
- \begin{frame}{Funktionen}
- \begin{figure}[h]
- \centering
- \includegraphics*[width=0.7\linewidth, keepaspectratio]{images/function-machine.png}
- \end{figure}
- \end{frame}
- \begin{frame}{Funktionen}
- \medskip
- \begin{columns}[t,onlytextwidth]
- \begin{column}{.5\textwidth}{
- \begin{itemize}[<+->]
- \item $f(x) = x^2$ ist $f: \mathbb{R} \rightarrow \mathbb{R}$
- \item $f(x, y) = x^2 + y^2$ ist $f: \mathbb{R}^2 \rightarrow \mathbb{R}$
- \item $f(x, y) = (x^2 + y^2, x \cdot y)$ ist $f: \mathbb{R}^2 \rightarrow \mathbb{R}^2$
- \end{itemize}
- }
- \end{column}
- \begin{column}{.4\textwidth}
- \only<1>{
- \begin{tikzpicture}
- \begin{axis}[
- legend pos=south west,
- axis x line=middle,
- axis y line=middle,
- grid = major,
- width=6.5cm,
- height=6.5cm,
- grid style={dashed, gray!30},
- xmin=-2, % start the diagram at this x-coordinate
- xmax= 2, % end the diagram at this x-coordinate
- ymin=-0.25, % start the diagram at this y-coordinate
- ymax= 4.25, % end the diagram at this y-coordinate
- axis background/.style={fill=white},
- xlabel=$x \in \mathbb{R}$,
- ylabel=$f(x) \in \mathbb{R}$,
- %xticklabels={-2,-1.6,...,7},
- %yticklabels={-8,-7,...,8},
- tick align=outside,
- minor tick num=-3,
- enlargelimits=true,
- tension=0.08]
- \addplot[domain=-2:2, red, thick,samples=40] {x*x};
- \end{axis}
- \end{tikzpicture}
- }
- \only<2->{
- \pgfplotsset{
- colormap={whitered}{
- color(0cm)=(white);
- color(1cm)=(orange!75!red)
- }
- }
- \begin{tikzpicture}
- \begin{axis}[
- colormap name=whitered,
- width=5.5cm,
- height=5.5cm,
- view={340}{25},
- enlargelimits=false,
- grid=major,
- domain=-3:3,
- y domain=-3:3,
- samples=56, %57 : TeX capacity exceeded, sorry [main memory size=3000000].
- % see also http://tex.stackexchange.com/a/7954/5645
- xlabel=$x$,
- ylabel=$y$,
- zlabel={$f(x,y)$},
- ]
- \addplot3[surf] {x^2 + y^2};
- \end{axis}
- \end{tikzpicture}
- }
- \end{column}
- \end{columns}
- \end{frame}
- \begin{frame}{Funktionen mit Parametern}
- \begin{columns}
- \begin{column}{.5\textwidth}
- {\Large Mit Parametern}
- \begin{itemize}[<+->]
- \item $f(x) = x^2$
- \item $f(x) = 2 \cdot x^2$
- \item $f(x) = \nicefrac{1}{2} \cdot x^2$
- \item $f(x) = a \cdot x^2$
- \item $f(x_1, \dots, x_{166}) = \sum_{i=1}^{166} a_i \cdot x_i$\\
- $\mathbb{R}^{166} \rightarrow \mathbb{R}$
- \item $f(x_1, \dots, x_{166}) = (\sum_{i=1}^{166} a_i \cdot x_i, \dots, \sum_{i=1}^{166} z_i \cdot x_i)$\\
- $\mathbb{R}^{166} \rightarrow \mathbb{R}^{\text(\# Klassen)}$
- \end{itemize}
- \end{column}
- \begin{column}{.4\textwidth}
- \begin{tikzpicture}
- \begin{axis}[
- legend pos=south west,
- axis x line=middle,
- axis y line=middle,
- grid = major,
- width=6.5cm,
- height=6.5cm,
- grid style={dashed, gray!30},
- xmin=-2, % start the diagram at this x-coordinate
- xmax= 2, % end the diagram at this x-coordinate
- ymin=-0.25, % start the diagram at this y-coordinate
- ymax= 4.25, % end the diagram at this y-coordinate
- axis background/.style={fill=white},
- xlabel=$x \in \mathbb{R}$,
- ylabel=$f(x) \in \mathbb{R}$,
- %xticklabels={-2,-1.6,...,7},
- %yticklabels={-8,-7,...,8},
- tick align=outside,
- minor tick num=-3,
- enlargelimits=true,
- tension=0.08]
- \only<1->{\addplot[domain=-2:2, red, thick,samples=40] {x*x};}
- \only<2->{\addplot[domain=-2:2, blue, thick,samples=40] {2*x*x};}
- \only<3->{\addplot[domain=-2:2, green, thick,samples=40] {0.5*x*x};}
- \end{axis}
- \end{tikzpicture}
- \end{column}
- \end{columns}
- \end{frame}
- \begin{frame}{Fehlerfunktion}
- \begin{itemize}[<+->]
- \item \textbf{Daten $(x_{i,1}, \dots, x_{i,n}, y_{i,1}, \dots, y_{i,\text{\# Klassen}})$}: Beispiele für den Computer
- \item \textbf{Aktuelles Modell $f$}: Funktion mit vielen Parametern
- \item \textbf{Fehlerfunktion}: Wie gut ist $f$ für die vorhandenen
- Daten?
- \end{itemize}
- \end{frame}
- \begin{frame}{Fehlerfunktion}
- Abbildung von \textbf{Parameterraum} auf den Fehler ($\mathbb{R}_0^+$)
- \end{frame}
- \begin{frame}{Minimieren mit Ableitungen}
- \begin{figure}[h]
- \centering
- \includegraphics*[width=0.7\linewidth, keepaspectratio]{images/derivative-function.png}
- \end{figure}
- \end{frame}
- \begin{frame}{Gradientenabstieg}
- \begin{figure}[h]
- \centering
- \includegraphics*[width=0.7\linewidth, keepaspectratio]{images/gradient-descent.png}
- \end{figure}
- \end{frame}
- \section{Neuronale Netze}
- \subsection{Neuronale Netze}
- \begin{frame}{Neuronale Netze}{}
- \begin{itemize}[<+->]
- \item Menge von parametrisierten Funktionen
- $\mathbb{R}^n \rightarrow \mathbb{R}^{\text(\# Klassen)}$
- \item $\mathbb{R}^n$: Eingabe,\\z.B. Farbe von Pixel~1, Farbe von Pixel~2, \dots
- \item $\mathbb{R}^{\text(\# Klassen)}$: Ausgabe,\\Wahrscheinlichkeit der Klasse (z.B. 0, 1, 2, 3, 4, 5, 6, 7, 8, 9)
- \item Ableitbar
- \end{itemize}
- \end{frame}
- \section{Ausblick}
- \subsection{Ausblick}
- \begin{frame}{Ausblick}
- Erkennung von Formeln
- \begin{itemize}[<+->]
- \item Aufbau eines Sprachmodells der Mathematik
- \item Erweiterung der Symboldatenbank
- \item Segmentierung
- \end{itemize}
- \end{frame}
|