📣 Post by dmm

2024-11-14 ┃ edited

Back when I was studying the Bloch Sphere (an incredibly beautiful way to describe the state of a qbit [1]), I noticed that the Kornecker/tensor product [2] of qubit states computes the one-hot representation of the n-bit state vector. These state vectors turn out to be the basis vectors in the complex space representing a quantum register. Anyway, this one-hot-like representation [3] reminded me of how inputs and labels for machine learning algorithms are frequently coded (at least in many supervised learning settings).

My question: Is there a deeper connection between the tensor products we find in quantum mechanics and the one-hot encodings we find in machine learning?

I wrote up a bit more, hoping to remember the details in the future. In any event, see the image (or scan Section 4 of davidmeyer.github.io/qc/bloch_…). As always, questions/comments/corrections/* greatly appreciated.