Verwendung von „if-else“ in neuronalen Netzwerken

Question

Ich würde gerne so zeichnen:

Derzeit habe ich

Mein aktueller Code ist

\begin{figure}
\centering
\label{fig:nn2}
\begin{tikzpicture}[shorten >=1pt,->,draw=black!50, node distance=\layersep]
    \tikzstyle{every pin edge}=[<-,shorten <=1pt]
    \tikzstyle{neuron}=[circle,fill=black!25,minimum size=17pt,inner sep=0pt]
    \tikzstyle{input neuron}=[neuron, fill=green!50];
    \tikzstyle{output neuron}=[neuron, fill=red!50];
    \tikzstyle{hidden neuron}=[neuron, fill=blue!50];
    \tikzstyle{annot} = [text width=4em, text centered]

    % Draw the input layer nodes
    \foreach \name / \y in {1,...,4}
    % This is the same as writing \foreach \name / \y in {1/1,2/2,3/3,4/4}
        %\if \y in {1,2,3,4}        
        \node[input neuron, pin=left:Input \#\y] (I-\name) at (0,-\y) {$x_{\name}$};
        %\else
        %\node[input neuron, pin=left:Input \#\y] (I-\name) at (0,-\y) {$x_{\name}$};
    %\node[input neuron, pin=left:Input \#4] (+1) at (0,-4) {$x_{\name}$};

    % Draw the hidden layer nodes
    \foreach \name / \y in {1,...,4}
        \path[yshift=0.5cm]
            node[hidden neuron] (H-\name) at (\layersep,-\y cm) {$h_{\name}$};

    % Draw the output layer node
    \node[output neuron,pin={[pin edge={->}]right:Output}, right of=H-3] (O) {$y_{0}$};

    % Connect every node in the input layer with every node in the
    % hidden layer.
    \foreach \source in {1,...,4}
        \foreach \dest in {1,...,4}
            \path (I-\source) edge (H-\dest);

    % Connect every node in the hidden layer with the output layer
    \foreach \source in {1,...,4}
        \path (H-\source) edge (O);

    % Annotate the layers
    \node[annot,above of=H-1, node distance=1cm] (hl) {Hidden layer};
    \node[annot,left of=hl] {Input layer};
    \node[annot,right of=hl] {Output layer};
\end{tikzpicture}
\caption{A figure shows the structure of a general neural networks model}
\end{figure}

Ich versuche, „if-else“ zu verwenden. und nach der Überarbeitung lautet mein Code:

\begin{figure}
\centering
\label{fig:nn2}
\begin{tikzpicture}[shorten >=1pt,->,draw=black!50, node distance=\layersep]
    \tikzstyle{every pin edge}=[<-,shorten <=1pt]
    \tikzstyle{neuron}=[circle,fill=black!25,minimum size=17pt,inner sep=0pt]
    \tikzstyle{input neuron}=[neuron, fill=green!50];
    \tikzstyle{output neuron}=[neuron, fill=red!50];
    \tikzstyle{hidden neuron}=[neuron, fill=blue!50];
    \tikzstyle{annot} = [text width=4em, text centered]

    % Draw the input layer nodes
    \foreach \name / \y in {1,...,4}
    % This is the same as writing \foreach \name / \y in {1/1,2/2,3/3,4/4}
        \if \y in {1,2,3,4}        
        \node[input neuron, pin=left:Input \#\y] (I-\name) at (0,-\y) {$x_{\name}$};
        \else
        %\node[input neuron, pin=left:Input \#\y] (I-\name) at (0,-\y) {$x_{\name}$};
    %\node[input neuron, pin=left:Input \#4] (+1) at (0,-4) {$x_{\name}$};

    % Draw the hidden layer nodes
    \foreach \name / \y in {1,...,4}
        \path[yshift=0.5cm]
            node[hidden neuron] (H-\name) at (\layersep,-\y cm) {$h_{\name}$};

    % Draw the output layer node
    \node[output neuron,pin={[pin edge={->}]right:Output}, right of=H-3] (O) {$y_{0}$};

    % Connect every node in the input layer with every node in the
    % hidden layer.
    \foreach \source in {1,...,4}
        \foreach \dest in {1,...,4}
            \path (I-\source) edge (H-\dest);

    % Connect every node in the hidden layer with the output layer
    \foreach \source in {1,...,4}
        \path (H-\source) edge (O);

    % Annotate the layers
    \node[annot,above of=H-1, node distance=1cm] (hl) {Hidden layer};
    \node[annot,left of=hl] {Input layer};
    \node[annot,right of=hl] {Output layer};
\end{tikzpicture}
\caption{A figure shows the structure of a general neural networks model}
\end{figure}

jedoch liegt ein Fehler vor: Extra }, oder vergessen \endgroup

Danke schön!

Answer 1

Es fehlten die geschweiften Klammern um den Schleifenkörper und die \fizum Abschließen der if-Anweisung. Außerdem \iffunktioniert die Anweisung nicht wie beabsichtigt:

\if<token1><token2>(Test, ob Zeichencodes übereinstimmen)

TeX erweitert die folgenden Makros, \ifbis zwei nicht erweiterbare Token gefunden werden. Wenn eines der Token eine Steuersequenz ist, betrachtet TeX es als Zeichencode 256 und Kategoriecode 16, es sei denn, das aktuelle Äquivalent dieser Steuersequenz war \letgleich einem nicht aktiven Zeichentoken. Auf diese Weise gibt jedes Token ein (Zeichencode, Kategoriecode)-Paar an. Die Bedingung ist erfüllt, wenn die Zeichencodes gleich sind, unabhängig von den Kategoriecodes. Beispielsweise sind nach \def\a{*}und \let\b=*und \def\c{/}die Tests \if*\aund \if\a\bwahr, aber \if\a\cfalsch. Also \if\a\parist falsch, aber \if\par\letist wahr.

(Das TeXbook Seite 209)

Daher vergleichst du im ersten Iterationsschritt 1und i, im zweiten 2und iso weiter und wertest immer false aus. Stattdessen prüfe ich, ob es der letzte Iterationsschritt ist mit \ifnum:

\begin{figure}
\centering
\label{fig:nn2}
\begin{tikzpicture}[shorten >=1pt,->,draw=black!50, node distance=\layersep]
    \tikzstyle{every pin edge}=[<-,shorten <=1pt]
    \tikzstyle{neuron}=[circle,fill=black!25,minimum size=17pt,inner sep=0pt]
    \tikzstyle{input neuron}=[neuron, fill=green!50];
    \tikzstyle{output neuron}=[neuron, fill=red!50];
    \tikzstyle{hidden neuron}=[neuron, fill=blue!50];
    \tikzstyle{annot} = [text width=4em, text centered]
    \newcommand{\n}{4} % number of neurons per layer

    % Draw the input layer nodes
    \foreach \name / \y in {1,...,\n}{
    % This is the same as writing \foreach \name / \y in {1/1,2/2,3/3,4/4}
        \ifnum \y=\n
            \node[input neuron, pin=left:Input \#$n$] (I-\name) at (0,-\y) {$x_{n}$};
        \else
            \node[input neuron, pin=left:Input \#\y] (I-\name) at (0,-\y) {$x_{\name}$};
        \fi
    }

    % Draw the hidden layer nodes
    \foreach \name / \y in {1,...,\n}{
        \ifnum \y=\n
            \path[yshift=0.5cm] node[hidden neuron] (H-\name) at (\layersep,-\y cm) {$h_{n}$};
        \else
            \path[yshift=0.5cm] node[hidden neuron] (H-\name) at (\layersep,-\y cm) {$h_{\name}$};
        \fi
    }

    % Draw the output layer node
    \node[output neuron,pin={[pin edge={->}]right:Output}, right of=H-3] (O) {$y_{0}$};

    % Connect every node in the input layer with every node in the
    % hidden layer.
    \foreach \source in {1,...,\n}
        \foreach \dest in {1,...,\n}
            \path (I-\source) edge (H-\dest);

    % Connect every node in the hidden layer with the output layer
    \foreach \source in {1,...,\n}
        \path (H-\source) edge (O);

    % Annotate the layers
    \node[annot,above of=H-1, node distance=1cm] (hl) {Hidden layer};
    \node[annot,left of=hl] {Input layer};
    \node[annot,right of=hl] {Output layer};
\end{tikzpicture}
\caption{A figure shows the structure of a general neural networks model}
\end{figure}

Allerdings würde ich hier kein if verwenden:

\begin{figure}
\centering
\label{fig:nn2}
\begin{tikzpicture}[shorten >=1pt,->,draw=black!50, node distance=\layersep]
    \tikzstyle{every pin edge}=[<-,shorten <=1pt]
    \tikzstyle{neuron}=[circle,fill=black!25,minimum size=17pt,inner sep=0pt]
    \tikzstyle{input neuron}=[neuron, fill=green!50];
    \tikzstyle{output neuron}=[neuron, fill=red!50];
    \tikzstyle{hidden neuron}=[neuron, fill=blue!50];
    \tikzstyle{annot} = [text width=4em, text centered]
    \newcommand{\numberNeuronsPerLayer}{4}
    \edef\numberNeuronsPerLayerMinusOne{\number\numexpr\numberNeuronsPerLayer-1\relax}

    % Draw the input layer nodes
    \foreach \name / \y in {1,...,\numberNeuronsPerLayerMinusOne}{
    % This is the same as writing \foreach \name / \y in {1/1,2/2,3/3,4/4}
        \node[input neuron, pin=left:Input \#\y] (I-\name) at (0,-\y) {$x_{\name}$};
    }
    \node[input neuron, pin=left:Input \#$n$] (I-\numberNeuronsPerLayer) at (0,-\numberNeuronsPerLayer) {$x_{n}$};

    % Draw the hidden layer nodes
    \begin{scope}[yshift=0.5cm]
        \foreach \name / \y in {1,...,\numberNeuronsPerLayerMinusOne}{
            \path node[hidden neuron] (H-\name) at (\layersep,-\y cm) {$h_{\name}$};
        }
        \path node[hidden neuron] (H-\numberNeuronsPerLayer) at (\layersep,-\numberNeuronsPerLayer cm) {$h_{n}$};
    \end{scope}

    % Draw the output layer node
    \node[output neuron,pin={[pin edge={->}]right:Output}, right of=H-3] (O) {$y_{0}$};

    % Connect every node in the input layer with every node in the
    % hidden layer.
    \foreach \source in {1,...,\numberNeuronsPerLayer}
        \foreach \dest in {1,...,\numberNeuronsPerLayer}
            \path (I-\source) edge (H-\dest);

    % Connect every node in the hidden layer with the output layer
    \foreach \source in {1,...,\numberNeuronsPerLayer}
        \path (H-\source) edge (O);

    % Annotate the layers
    \node[annot,above of=H-1, node distance=1cm] (hl) {Hidden layer};
    \node[annot,left of=hl] {Input layer};
    \node[annot,right of=hl] {Output layer};
\end{tikzpicture}
\caption{A figure shows the structure of a general neural networks model}
\end{figure}

Answer 2

Basierend auf meiner Antwort auf Ihre Frage:

\documentclass[tikz, margin=3mm]{standalone}
\usetikzlibrary{calc, chains, positioning}

\begin{document}
    \begin{tikzpicture}[shorten >=1pt,->, draw=black!50,
        node distance = 6mm and 24mm,
          start chain = going below,
every pin edge/.style = {<-,shorten <=1pt},
        neuron/.style = {circle, fill=#1,
                         minimum size=17pt, inner sep=0pt},
         annot/.style = {text width=4em, align=center}
                        ]
% Draw the input and hidden layer nodes
\ifnum\i<4 
    \node[neuron=green!50, on chain,
          pin=180:Input \#\i    % if you not like to have this inputs, just erase them
          ]              (I-\i)  {$x_{\i}$};
    \node[neuron=blue!50,
      right=of I-\i]     (H-\i)  {};
\else 
    \node[neuron=green!50, on chain,
          pin=180:Input \#\i    % if you not like to have this inputs, just erase them
          ]              (I-\i)  {$+1$}; 
    \node[neuron=blue!50,
          right=of I-\i] (H-\i)  {$+1$}; 
\fi
}
% Draw the output layer node
    \node[neuron=red!50,
          right=of $(H-2)!0.5!(H-3)$]  (O-1)   {};
% Connect input nodes with hidden nodes and
%  hiden nodes with output nodes with the output layer
    \foreach \i in {1,...,4}
        \foreach \j in {1,...,4}
{
    \draw (I-\i) edge (H-\j)
          (H-\j) edge (O-1);
}
    \draw (O-1) -- node[below] {$h_{w,b}(x)$} + (2,0);
% Annotate layers
\node[annot,below=of I-4.center]        {Layer 1};
\node[annot,below=of H-4.center]        {Layer 2};
\node[annot,below=of O-1 |- H-4.center] {Layer 3};
    \end{tikzpicture}
\end{document}

Konditionalsätze: wie aus obigem MWE folgt

\ifnum\i<4
    action 1
\else
    action 2
\fi;

Nachtrag: Ihr Problem können Sie ohne bedingte Anweisung lösen:

\documentclass[tikz, margin=3mm]{standalone}
\usetikzlibrary{calc, chains, positioning}

\begin{document}
    \begin{tikzpicture}[shorten >=1pt,->, draw=black!50,
        node distance = 6mm and 24mm,
          start chain = going below,
every pin edge/.style = {<-,shorten <=1pt},
        neuron/.style = {circle, fill=#1,
                         minimum size=17pt, inner sep=0pt},
         annot/.style = {text width=4em, align=center}
                        ]
% Draw the input and hyden layer nodes
\foreach \i in {1,2,3}
{
    \node[neuron=green!50, on chain,
          pin=180:Input \#\i    % if you not like to have this inputs, just erase them
          ]             (I-\i)  {$x_{\i}$};
    \node[neuron=blue!50,
      right=of I-\i]    (H-\i)     {};
}
    \node[neuron=green!50, 
          pin=180:Input \#\i    % if you not like to have this inputs, just erase them
          below=of I-3
          ]              (I-4)  {$+1$};
    \node[neuron=blue!50,
          below=of H-3] (H-4)  {$+1$};
% Draw the output layer node
    \node[neuron=red!50,
          right=of $(H-2)!0.5!(H-3)$]  (O-1)   {};
% Connect input nodes with hidden nodes and
%  hiden nodes with output nodes with the output layer
    \foreach \i in {1,...,4}
        \foreach \j in {1,...,4}
{
    \draw (I-\i) edge (H-\j)
          (H-\j) edge (O-1);
}
    \draw (O-1) -- node[below] {$h_{w,b}(x)$} + (2,0);
% Annotate layers
\node[annot,below=of I-4.center]        {Layer 1};
\node[annot,below=of H-4.center]        {Layer 2};
\node[annot,below=of O-1 |- H-4.center] {Layer 3};
    \end{tikzpicture}
\end{document}

Das Ergebnis ist das gleiche wie zuvor.