This is an attempt to convert online version of Michael Nielsen's book 'Neural Networks and Deep Learning' into LaTeX source.
- Chapter 1: done
- Chapter 2: done
- Chapter 3: done
- Chapter 4: includes a lot of interactive JS-based elements. In progress. By now, interactive elements are replaced with intuitive (I hope) graphs, but text is not fully adapted.
- Chapter 5: done
- Chapter 6: done
I observed some missed Python code in the online version of network3.py:
print('The corresponding test accuracy is {0:.2
test_accuracy))
...
print("Best validation accuracy of {0:.2
best_validation_accuracy, best_iteration))
...
print("Corresponding test accuracy of {0:.2
So, these parts were replaced with the correct ones from the source code repo.
Can be compiled into any desired format, using XeLaTeX — with any desired font.
As a general design, I used my PhD thesis style: 17x24 cm paper, 9pt font, Charter/Mathdesign, own designed charper titles, chapter labels etc.
Typography adjusted (- → –, "" → “ ”)
Bibliography — maybe to collect all cited research papers?
Equation numbering is updated to sequential as in the original online book. Please note that some numbers are missing (e.g. 40-41), since some equations in the online book are multiline with a label on every line. I use the same tags/numbers as in the book.
Epub version added.
pandoc -s --mathml book.tex -o book.epub
converts source latex files into epub with formulas redneder by MathML. MathML works correctly in Calibre.
Please note: pandoc does not produce images from tikzpicture, therefore chapter 4 in epub is corrupted with missing images. It in much better to check Chapter 4 online anyway, since it contains interactive elements.