Most of our static diagrams are drawn in a vector graphics tool, like Adobe Illustrator, Sketch or Inkscape. For dynamic diagrams we generally use D3.js.
Absolutely! We’ve put a lot of work into making the most helpful diagrams we can, and we’d be delighted for them to be useful to others.
You are welcome to use our diagrams in your talk/paper/post/etc under the Creative Commons Attribution License. The javascript code required to reproduce the interactive diagrams is released under an MIT license, unless noted otherwise.
If you use our diagrams, we do ask that you give clear attribution. In an academic context, we recommend you include a citation in the caption for the diagram. (You can find bibtex for citing our articles in a link at the top of the article.)
Please note that some of our posts include diagrams borrowed from papers that we’re explaining. You can recognize these figures because they have captions like “Figure from XYZ et al…” Since these diagrams aren’t ours, we can’t give you permission to use them.
That would be wonderful! We only ask that you make it clear that your translation is based on our post. Please let us know about any translations you write so we can link to them from the post! :)
No, sorry. We recommend you ask your friends, colleagues, teachers, or an appropriate online community. It’s not unusual for us to get dozens of emails asking for help with different things every week. We wish we could help all of them, but we think it’s better for us to make awesome articles that help hundreds of thousands of people.
Michael Nielsen’s free online book, Neural Networks and Deep Learning, is an outstanding place to start. It introduces a lot of core ideas in a very approachable way. For a more comprehensive resource, we recommend you look at Goodfellow, Bengio & Courville’s Deep Learning book.
Other excellent resources include the TensorFlow Tutorials, Andrej Karpathy’s blog, and Chris Olah’s blog.