I started writing this book teaching tensor diagrams, which are a notation from Physics for working with tensors (think 3d matrices and beyond).
Oftentimes math notation, in particular in machine learning papers, need to work with tensors, but lack sufficient notation, and easy things, like taking derivatives, becomes hard.
I'm basically rewriting "the matrix cookbook" using tensors, and I think it makes everything much more intuitive. It's still just a draft, but I would love your feedback!
1 comment
[ 2.8 ms ] story [ 12.2 ms ] threadOftentimes math notation, in particular in machine learning papers, need to work with tensors, but lack sufficient notation, and easy things, like taking derivatives, becomes hard.
I'm basically rewriting "the matrix cookbook" using tensors, and I think it makes everything much more intuitive. It's still just a draft, but I would love your feedback!