MathematicalEngineeringDeepLearning/tikz/figure5-2-a-realvalue_function_approximation.tex at master · yoninazarathy/MathematicalEngineeringDeepLearning · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
\documentclass[tikz, border=50pt]{standalone}

\usepackage{tikz}
\usepackage{medl_colors}
\usetikzlibrary{shapes.multipart, shapes.geometric, arrows.meta}
\usetikzlibrary{matrix, calc, positioning,fit}
\usepackage{arrayjob}

\begin {document}

\def\negationorder{{{"1","2","r-1"},{"0","1","r-2"}}}

\begin{tikzpicture}

% Layer A
\node[greenshape, circle, minimum size = 1cm] (layerA0) at (0,-5) {\large {$x$}};

% Layer B
\node[blueshape, circle, minimum size = 2.5cm] (layerB0) at (6, 0) {bias = $-u_1$};
\node[blueshape, circle, minimum size = 2.5cm] (layerB1) at (6, -3) {bias = $-u_2$};
\node[blueshape, circle, minimum size = 2.5cm] (layerB2) at (6, -10) {bias = $-u_{r-1}$};

%layer C
\draw [thick, dashed] ([yshift=-.3cm]layerB1.south) -- ([yshift=.3cm]layerB2.north);
\node[redshape, circle, minimum size = 1cm, align=center] (layerC0) at (12, -5) {\large {$\hat y$}};

%connecting between the nodes
\foreach \i in {0,1,2}
{
      \pgfmathparse{\negationorder[0][\i]} \edef\lsign{\pgfmathresult};
      \pgfmathparse{\negationorder[1][\i]} \edef\rsign{\pgfmathresult};

       \draw[-Triangle, thickline] (layerA0) -- (layerB\i) node[pos=.5, above, sloped] {$w^{[1]}_{1,\lsign}=1$};
       \draw[-Triangle, thickline] (layerB\i) -- (layerC0) node[pos=.5, above, sloped] {$w^{[2]}_{\lsign,1}=s_{\lsign}-s_{\rsign}$};
}

\end{tikzpicture}
\end {document}