Skip to content

Commit 8425e0b

Browse files
mhjensenmhjensen
committed
minor update
1 parent 5a3e6d2 commit 8425e0b

File tree

139 files changed

+34986
-26
lines changed

Some content is hidden

Large Commits have some content hidden by default. Use the searchbox below for content that may be hidden.

139 files changed

+34986
-26
lines changed

‎doc/pub/week16/html/week16-bs.html‎

Lines changed: 991 additions & 4 deletions
Large diffs are not rendered by default.

‎doc/pub/week16/html/week16-reveal.html‎

Lines changed: 979 additions & 3 deletions
Large diffs are not rendered by default.

‎doc/pub/week16/html/week16-solarized.html‎

Lines changed: 961 additions & 4 deletions
Large diffs are not rendered by default.

‎doc/pub/week16/html/week16.html‎

Lines changed: 961 additions & 4 deletions
Large diffs are not rendered by default.

‎doc/pub/week16/ipynb/ipynb-week16-src.tar.gz‎

-2 Bytes
Binary file not shown.

‎doc/pub/week16/ipynb/week16.ipynb‎

Lines changed: 1570 additions & 6 deletions
Large diffs are not rendered by default.

‎doc/pub/week16/pdf/week16.pdf‎

268 KB
Binary file not shown.

‎doc/src/week16/README.txt‎

Lines changed: 2 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,2 @@
1+
This IPython notebook week16.ipynb does not require any additional
2+
programs.

‎doc/src/week16/_minted-week16/0B3C199E80015739FC2F8AC85DC873A993D35B036401E9069075B6880EFF14DD.pygtex‎

Lines changed: 84 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,84 @@
1+
\begin{Verbatim}[commandchars=\\\{\},codes={\catcode`\$=3\catcode`\^=7\catcode`\_=8\relax}]
2+
\PYG{k+kn}{import} \PYG{n+nn}{numpy} \PYG{k}{as} \PYG{n+nn}{np}
3+
\PYG{k+kn}{from} \PYG{n+nn}{sklearn.datasets} \PYG{k+kn}{import} \PYG{n}{fetch\PYGZus{}openml}
4+
\PYG{k+kn}{from} \PYG{n+nn}{sklearn.model\PYGZus{}selection} \PYG{k+kn}{import} \PYG{n}{train\PYGZus{}test\PYGZus{}split}
5+
\PYG{k+kn}{import} \PYG{n+nn}{matplotlib.pyplot} \PYG{k}{as} \PYG{n+nn}{plt}
6+
7+
\PYG{c+c1}{\PYGZsh{} Load and preprocess MNIST}
8+
\PYG{k}{def} \PYG{n+nf}{load\PYGZus{}binarized\PYGZus{}mnist}\PYG{p}{():}
9+
\PYG{n+nb}{print}\PYG{p}{(}\PYG{l+s+s2}{\PYGZdq{}Downloading MNIST...\PYGZdq{}}\PYG{p}{)}
10+
\PYG{n}{mnist} \PYG{o}{=} \PYG{n}{fetch\PYGZus{}openml}\PYG{p}{(}\PYG{l+s+s1}{\PYGZsq{}mnist\PYGZus{}784\PYGZsq{}}\PYG{p}{,} \PYG{n}{version}\PYG{o}{=}\PYG{l+m+mi}{1}\PYG{p}{)}
11+
\PYG{n}{X} \PYG{o}{=} \PYG{n}{mnist}\PYG{o}{.}\PYG{n}{data}\PYG{o}{.}\PYG{n}{astype}\PYG{p}{(}\PYG{n}{np}\PYG{o}{.}\PYG{n}{float32}\PYG{p}{)} \PYG{o}{/} \PYG{l+m+mf}{255.0}
12+
\PYG{n}{X} \PYG{o}{=} \PYG{p}{(}\PYG{n}{X} \PYG{o}{\PYGZgt{}} \PYG{l+m+mf}{0.5}\PYG{p}{)}\PYG{o}{.}\PYG{n}{astype}\PYG{p}{(}\PYG{n}{np}\PYG{o}{.}\PYG{n}{float32}\PYG{p}{)} \PYG{c+c1}{\PYGZsh{} Binarize}
13+
\PYG{k}{return} \PYG{n}{X}
14+
15+
\PYG{k}{class} \PYG{n+nc}{RBM}\PYG{p}{:}
16+
\PYG{k}{def} \PYG{n+nf+fm}{\PYGZus{}\PYGZus{}init\PYGZus{}\PYGZus{}}\PYG{p}{(}\PYG{n+nb+bp}{self}\PYG{p}{,} \PYG{n}{n\PYGZus{}visible}\PYG{p}{,} \PYG{n}{n\PYGZus{}hidden}\PYG{p}{,} \PYG{n}{learning\PYGZus{}rate}\PYG{o}{=}\PYG{l+m+mf}{0.1}\PYG{p}{):}
17+
\PYG{n+nb+bp}{self}\PYG{o}{.}\PYG{n}{n\PYGZus{}visible} \PYG{o}{=} \PYG{n}{n\PYGZus{}visible}
18+
\PYG{n+nb+bp}{self}\PYG{o}{.}\PYG{n}{n\PYGZus{}hidden} \PYG{o}{=} \PYG{n}{n\PYGZus{}hidden}
19+
\PYG{n+nb+bp}{self}\PYG{o}{.}\PYG{n}{learning\PYGZus{}rate} \PYG{o}{=} \PYG{n}{learning\PYGZus{}rate}
20+
21+
\PYG{c+c1}{\PYGZsh{} Initialize weights and biases}
22+
\PYG{n+nb+bp}{self}\PYG{o}{.}\PYG{n}{W} \PYG{o}{=} \PYG{n}{np}\PYG{o}{.}\PYG{n}{random}\PYG{o}{.}\PYG{n}{normal}\PYG{p}{(}\PYG{l+m+mi}{0}\PYG{p}{,} \PYG{l+m+mf}{0.01}\PYG{p}{,} \PYG{n}{size}\PYG{o}{=}\PYG{p}{(}\PYG{n}{n\PYGZus{}visible}\PYG{p}{,} \PYG{n}{n\PYGZus{}hidden}\PYG{p}{))}
23+
\PYG{n+nb+bp}{self}\PYG{o}{.}\PYG{n}{v\PYGZus{}bias} \PYG{o}{=} \PYG{n}{np}\PYG{o}{.}\PYG{n}{zeros}\PYG{p}{(}\PYG{n}{n\PYGZus{}visible}\PYG{p}{)}
24+
\PYG{n+nb+bp}{self}\PYG{o}{.}\PYG{n}{h\PYGZus{}bias} \PYG{o}{=} \PYG{n}{np}\PYG{o}{.}\PYG{n}{zeros}\PYG{p}{(}\PYG{n}{n\PYGZus{}hidden}\PYG{p}{)}
25+
26+
\PYG{k}{def} \PYG{n+nf}{sigmoid}\PYG{p}{(}\PYG{n+nb+bp}{self}\PYG{p}{,} \PYG{n}{x}\PYG{p}{):}
27+
\PYG{k}{return} \PYG{l+m+mi}{1} \PYG{o}{/} \PYG{p}{(}\PYG{l+m+mi}{1} \PYG{o}{+} \PYG{n}{np}\PYG{o}{.}\PYG{n}{exp}\PYG{p}{(}\PYG{o}{\PYGZhy{}}\PYG{n}{x}\PYG{p}{))}
28+
29+
\PYG{k}{def} \PYG{n+nf}{sample}\PYG{p}{(}\PYG{n+nb+bp}{self}\PYG{p}{,} \PYG{n}{probs}\PYG{p}{):}
30+
\PYG{k}{return} \PYG{p}{(}\PYG{n}{np}\PYG{o}{.}\PYG{n}{random}\PYG{o}{.}\PYG{n}{rand}\PYG{p}{(}\PYG{o}{*}\PYG{n}{probs}\PYG{o}{.}\PYG{n}{shape}\PYG{p}{)} \PYG{o}{\PYGZlt{}} \PYG{n}{probs}\PYG{p}{)}\PYG{o}{.}\PYG{n}{astype}\PYG{p}{(}\PYG{n}{np}\PYG{o}{.}\PYG{n}{float32}\PYG{p}{)}
31+
32+
\PYG{k}{def} \PYG{n+nf}{train}\PYG{p}{(}\PYG{n+nb+bp}{self}\PYG{p}{,} \PYG{n}{data}\PYG{p}{,} \PYG{n}{epochs}\PYG{o}{=}\PYG{l+m+mi}{10}\PYG{p}{,} \PYG{n}{batch\PYGZus{}size}\PYG{o}{=}\PYG{l+m+mi}{64}\PYG{p}{):}
33+
\PYG{n}{n\PYGZus{}samples} \PYG{o}{=} \PYG{n}{data}\PYG{o}{.}\PYG{n}{shape}\PYG{p}{[}\PYG{l+m+mi}{0}\PYG{p}{]}
34+
\PYG{c+c1}{\PYGZsh{} Convert the DataFrame to a NumPy array to avoid the KeyError.}
35+
\PYG{n}{data} \PYG{o}{=} \PYG{n}{data}\PYG{o}{.}\PYG{n}{to\PYGZus{}numpy}\PYG{p}{()}
36+
\PYG{k}{for} \PYG{n}{epoch} \PYG{o+ow}{in} \PYG{n+nb}{range}\PYG{p}{(}\PYG{n}{epochs}\PYG{p}{):}
37+
\PYG{n}{np}\PYG{o}{.}\PYG{n}{random}\PYG{o}{.}\PYG{n}{shuffle}\PYG{p}{(}\PYG{n}{data}\PYG{p}{)}
38+
\PYG{n}{epoch\PYGZus{}error} \PYG{o}{=} \PYG{l+m+mi}{0}
39+
40+
\PYG{k}{for} \PYG{n}{i} \PYG{o+ow}{in} \PYG{n+nb}{range}\PYG{p}{(}\PYG{l+m+mi}{0}\PYG{p}{,} \PYG{n}{n\PYGZus{}samples}\PYG{p}{,} \PYG{n}{batch\PYGZus{}size}\PYG{p}{):}
41+
\PYG{n}{v0} \PYG{o}{=} \PYG{n}{data}\PYG{p}{[}\PYG{n}{i}\PYG{p}{:}\PYG{n}{i} \PYG{o}{+} \PYG{n}{batch\PYGZus{}size}\PYG{p}{]}
42+
\PYG{n}{h0\PYGZus{}prob} \PYG{o}{=} \PYG{n+nb+bp}{self}\PYG{o}{.}\PYG{n}{sigmoid}\PYG{p}{(}\PYG{n}{np}\PYG{o}{.}\PYG{n}{dot}\PYG{p}{(}\PYG{n}{v0}\PYG{p}{,} \PYG{n+nb+bp}{self}\PYG{o}{.}\PYG{n}{W}\PYG{p}{)} \PYG{o}{+} \PYG{n+nb+bp}{self}\PYG{o}{.}\PYG{n}{h\PYGZus{}bias}\PYG{p}{)}
43+
\PYG{n}{h0\PYGZus{}sample} \PYG{o}{=} \PYG{n+nb+bp}{self}\PYG{o}{.}\PYG{n}{sample}\PYG{p}{(}\PYG{n}{h0\PYGZus{}prob}\PYG{p}{)}
44+
45+
\PYG{n}{v1\PYGZus{}prob} \PYG{o}{=} \PYG{n+nb+bp}{self}\PYG{o}{.}\PYG{n}{sigmoid}\PYG{p}{(}\PYG{n}{np}\PYG{o}{.}\PYG{n}{dot}\PYG{p}{(}\PYG{n}{h0\PYGZus{}sample}\PYG{p}{,} \PYG{n+nb+bp}{self}\PYG{o}{.}\PYG{n}{W}\PYG{o}{.}\PYG{n}{T}\PYG{p}{)} \PYG{o}{+} \PYG{n+nb+bp}{self}\PYG{o}{.}\PYG{n}{v\PYGZus{}bias}\PYG{p}{)}
46+
\PYG{n}{h1\PYGZus{}prob} \PYG{o}{=} \PYG{n+nb+bp}{self}\PYG{o}{.}\PYG{n}{sigmoid}\PYG{p}{(}\PYG{n}{np}\PYG{o}{.}\PYG{n}{dot}\PYG{p}{(}\PYG{n}{v1\PYGZus{}prob}\PYG{p}{,} \PYG{n+nb+bp}{self}\PYG{o}{.}\PYG{n}{W}\PYG{p}{)} \PYG{o}{+} \PYG{n+nb+bp}{self}\PYG{o}{.}\PYG{n}{h\PYGZus{}bias}\PYG{p}{)}
47+
48+
\PYG{c+c1}{\PYGZsh{} Weight and bias updates}
49+
\PYG{n+nb+bp}{self}\PYG{o}{.}\PYG{n}{W} \PYG{o}{+=} \PYG{n+nb+bp}{self}\PYG{o}{.}\PYG{n}{learning\PYGZus{}rate} \PYG{o}{*} \PYG{p}{(}\PYG{n}{np}\PYG{o}{.}\PYG{n}{dot}\PYG{p}{(}\PYG{n}{v0}\PYG{o}{.}\PYG{n}{T}\PYG{p}{,} \PYG{n}{h0\PYGZus{}prob}\PYG{p}{)} \PYG{o}{\PYGZhy{}} \PYG{n}{np}\PYG{o}{.}\PYG{n}{dot}\PYG{p}{(}\PYG{n}{v1\PYGZus{}prob}\PYG{o}{.}\PYG{n}{T}\PYG{p}{,} \PYG{n}{h1\PYGZus{}prob}\PYG{p}{))} \PYG{o}{/} \PYG{n}{batch\PYGZus{}size}
50+
\PYG{n+nb+bp}{self}\PYG{o}{.}\PYG{n}{v\PYGZus{}bias} \PYG{o}{+=} \PYG{n+nb+bp}{self}\PYG{o}{.}\PYG{n}{learning\PYGZus{}rate} \PYG{o}{*} \PYG{n}{np}\PYG{o}{.}\PYG{n}{mean}\PYG{p}{(}\PYG{n}{v0} \PYG{o}{\PYGZhy{}} \PYG{n}{v1\PYGZus{}prob}\PYG{p}{,} \PYG{n}{axis}\PYG{o}{=}\PYG{l+m+mi}{0}\PYG{p}{)}
51+
\PYG{n+nb+bp}{self}\PYG{o}{.}\PYG{n}{h\PYGZus{}bias} \PYG{o}{+=} \PYG{n+nb+bp}{self}\PYG{o}{.}\PYG{n}{learning\PYGZus{}rate} \PYG{o}{*} \PYG{n}{np}\PYG{o}{.}\PYG{n}{mean}\PYG{p}{(}\PYG{n}{h0\PYGZus{}prob} \PYG{o}{\PYGZhy{}} \PYG{n}{h1\PYGZus{}prob}\PYG{p}{,} \PYG{n}{axis}\PYG{o}{=}\PYG{l+m+mi}{0}\PYG{p}{)}
52+
53+
\PYG{n}{epoch\PYGZus{}error} \PYG{o}{+=} \PYG{n}{np}\PYG{o}{.}\PYG{n}{mean}\PYG{p}{((}\PYG{n}{v0} \PYG{o}{\PYGZhy{}} \PYG{n}{v1\PYGZus{}prob}\PYG{p}{)} \PYG{o}{**} \PYG{l+m+mi}{2}\PYG{p}{)}
54+
55+
\PYG{n+nb}{print}\PYG{p}{(}\PYG{l+s+sa}{f}\PYG{l+s+s2}{\PYGZdq{}Epoch }\PYG{l+s+si}{\PYGZob{}}\PYG{n}{epoch}\PYG{+w}{ }\PYG{o}{+}\PYG{+w}{ }\PYG{l+m+mi}{1}\PYG{l+s+si}{\PYGZcb{}}\PYG{l+s+s2}{: Reconstruction error = }\PYG{l+s+si}{\PYGZob{}}\PYG{n}{epoch\PYGZus{}error}\PYG{l+s+si}{:}\PYG{l+s+s2}{.4f}\PYG{l+s+si}{\PYGZcb{}}\PYG{l+s+s2}{\PYGZdq{}}\PYG{p}{)}
56+
57+
\PYG{k}{def} \PYG{n+nf}{reconstruct}\PYG{p}{(}\PYG{n+nb+bp}{self}\PYG{p}{,} \PYG{n}{v}\PYG{p}{):}
58+
\PYG{n}{h} \PYG{o}{=} \PYG{n+nb+bp}{self}\PYG{o}{.}\PYG{n}{sigmoid}\PYG{p}{(}\PYG{n}{np}\PYG{o}{.}\PYG{n}{dot}\PYG{p}{(}\PYG{n}{v}\PYG{p}{,} \PYG{n+nb+bp}{self}\PYG{o}{.}\PYG{n}{W}\PYG{p}{)} \PYG{o}{+} \PYG{n+nb+bp}{self}\PYG{o}{.}\PYG{n}{h\PYGZus{}bias}\PYG{p}{)}
59+
\PYG{n}{v\PYGZus{}recon} \PYG{o}{=} \PYG{n+nb+bp}{self}\PYG{o}{.}\PYG{n}{sigmoid}\PYG{p}{(}\PYG{n}{np}\PYG{o}{.}\PYG{n}{dot}\PYG{p}{(}\PYG{n}{h}\PYG{p}{,} \PYG{n+nb+bp}{self}\PYG{o}{.}\PYG{n}{W}\PYG{o}{.}\PYG{n}{T}\PYG{p}{)} \PYG{o}{+} \PYG{n+nb+bp}{self}\PYG{o}{.}\PYG{n}{v\PYGZus{}bias}\PYG{p}{)}
60+
\PYG{k}{return} \PYG{n}{v\PYGZus{}recon}
61+
62+
\PYG{c+c1}{\PYGZsh{} Load and split MNIST}
63+
\PYG{n}{X} \PYG{o}{=} \PYG{n}{load\PYGZus{}binarized\PYGZus{}mnist}\PYG{p}{()}
64+
\PYG{n}{X\PYGZus{}train}\PYG{p}{,} \PYG{n}{X\PYGZus{}test} \PYG{o}{=} \PYG{n}{train\PYGZus{}test\PYGZus{}split}\PYG{p}{(}\PYG{n}{X}\PYG{p}{,} \PYG{n}{test\PYGZus{}size}\PYG{o}{=}\PYG{l+m+mf}{0.1}\PYG{p}{,} \PYG{n}{random\PYGZus{}state}\PYG{o}{=}\PYG{l+m+mi}{42}\PYG{p}{)}
65+
66+
\PYG{c+c1}{\PYGZsh{} Initialize and train RBM}
67+
\PYG{n}{rbm} \PYG{o}{=} \PYG{n}{RBM}\PYG{p}{(}\PYG{n}{n\PYGZus{}visible}\PYG{o}{=}\PYG{l+m+mi}{784}\PYG{p}{,} \PYG{n}{n\PYGZus{}hidden}\PYG{o}{=}\PYG{l+m+mi}{128}\PYG{p}{,} \PYG{n}{learning\PYGZus{}rate}\PYG{o}{=}\PYG{l+m+mf}{0.1}\PYG{p}{)}
68+
\PYG{n}{rbm}\PYG{o}{.}\PYG{n}{train}\PYG{p}{(}\PYG{n}{X\PYGZus{}train}\PYG{p}{,} \PYG{n}{epochs}\PYG{o}{=}\PYG{l+m+mi}{10}\PYG{p}{,} \PYG{n}{batch\PYGZus{}size}\PYG{o}{=}\PYG{l+m+mi}{64}\PYG{p}{)}
69+
70+
\PYG{c+c1}{\PYGZsh{} Visualize reconstruction}
71+
\PYG{k}{def} \PYG{n+nf}{show\PYGZus{}reconstruction}\PYG{p}{(}\PYG{n}{original}\PYG{p}{,} \PYG{n}{reconstructed}\PYG{p}{):}
72+
\PYG{n}{fig}\PYG{p}{,} \PYG{n}{axes} \PYG{o}{=} \PYG{n}{plt}\PYG{o}{.}\PYG{n}{subplots}\PYG{p}{(}\PYG{l+m+mi}{1}\PYG{p}{,} \PYG{l+m+mi}{2}\PYG{p}{)}
73+
\PYG{n}{axes}\PYG{p}{[}\PYG{l+m+mi}{0}\PYG{p}{]}\PYG{o}{.}\PYG{n}{imshow}\PYG{p}{(}\PYG{n}{original}\PYG{o}{.}\PYG{n}{reshape}\PYG{p}{(}\PYG{l+m+mi}{28}\PYG{p}{,} \PYG{l+m+mi}{28}\PYG{p}{),} \PYG{n}{cmap}\PYG{o}{=}\PYG{l+s+s2}{\PYGZdq{}gray\PYGZdq{}}\PYG{p}{)}
74+
\PYG{n}{axes}\PYG{p}{[}\PYG{l+m+mi}{0}\PYG{p}{]}\PYG{o}{.}\PYG{n}{set\PYGZus{}title}\PYG{p}{(}\PYG{l+s+s2}{\PYGZdq{}Original\PYGZdq{}}\PYG{p}{)}
75+
\PYG{n}{axes}\PYG{p}{[}\PYG{l+m+mi}{1}\PYG{p}{]}\PYG{o}{.}\PYG{n}{imshow}\PYG{p}{(}\PYG{n}{reconstructed}\PYG{o}{.}\PYG{n}{reshape}\PYG{p}{(}\PYG{l+m+mi}{28}\PYG{p}{,} \PYG{l+m+mi}{28}\PYG{p}{),} \PYG{n}{cmap}\PYG{o}{=}\PYG{l+s+s2}{\PYGZdq{}gray\PYGZdq{}}\PYG{p}{)}
76+
\PYG{n}{axes}\PYG{p}{[}\PYG{l+m+mi}{1}\PYG{p}{]}\PYG{o}{.}\PYG{n}{set\PYGZus{}title}\PYG{p}{(}\PYG{l+s+s2}{\PYGZdq{}Reconstruction\PYGZdq{}}\PYG{p}{)}
77+
\PYG{n}{plt}\PYG{o}{.}\PYG{n}{show}\PYG{p}{()}
78+
79+
\PYG{n}{sample} \PYG{o}{=} \PYG{n}{X\PYGZus{}test}\PYG{o}{.}\PYG{n}{iloc}\PYG{p}{[}\PYG{l+m+mi}{0}\PYG{p}{]}\PYG{o}{.}\PYG{n}{values} \PYG{c+c1}{\PYGZsh{} Access the first row and convert to NumPy array}
80+
\PYG{n}{reconstruction} \PYG{o}{=} \PYG{n}{rbm}\PYG{o}{.}\PYG{n}{reconstruct}\PYG{p}{(}\PYG{n}{sample}\PYG{p}{[}\PYG{n}{np}\PYG{o}{.}\PYG{n}{newaxis}\PYG{p}{,} \PYG{p}{:])}
81+
\PYG{n}{show\PYGZus{}reconstruction}\PYG{p}{(}\PYG{n}{sample}\PYG{p}{,} \PYG{n}{reconstruction}\PYG{p}{[}\PYG{l+m+mi}{0}\PYG{p}{])}
82+
83+
84+
\end{Verbatim}

0 commit comments

Comments
 (0)