Fantastic repository of notes on many many machine learning papers by Vitaly Kurin.

A great article showing how to perfectly model any dataset using a single parameter (there’s a lot of information hiding in that parameter). It is more of an intellectual exercise / thought experiment than an actual academic paper.

Tensorflow Lattice, which seems like a nice way of encoding real-world constraints into simple ML models.

A series of “pen and paper” exercises in machine learning: common computations and exercises that can all be done on paper, with solutions included. A really nice resource.

A nice paper about the difference between explanation and prediction.

A summary of some geometric, topological and algebraic structures associated with machine learning.