Credits

These notes draw mainly from two practitioner-authored books:

• Deep Learning: A Visual Approach (2021) by Andrew Glassner
• How AI Works: From Sorcery to Science (2023) by Ronald T. Kneusel

Both are grounded in influential research, including work published on arXiv.

Glassner's work stands out for its extensive use of illustrations to explain deep learning concepts.

This was particularly valuable to me: these diagrams provided an accessible way to build intuition.

As a result, the structure of these notes closely follows that of Glassner's book.

From over 800 figures, I selected around 300 and redrew many as SVGs.

In doing so, I used Anton Tsitsulin's 12-color qualitative (or categorical) palette and the Roboto typeface.

Both books are effective at building intuition without heavy math. That said, full understanding ultimately requires engaging with the underlying mathematics and algorithms.