[SomeStupidPoint]

One of the people mentioned in the article, Svore, works at the QuArC group for MS and has worked on quantum machine learning. (QuArC generally works on how to actually design and use quantum computers.)

For example: https://arxiv.org/pdf/1412.3489.pdf

From my (shallow) understanding, you can get improvements by considering the space of possible states of your quantum system to be the space of weights for your model, and then annealing the quantum system (which is equivalent to optimizing the objective function). The quantum annealing is a (significant) improvement over other optimization methods... or something.

[martinquantum]

+1

[martinquantum]

looks like a quantum computer can train models better. Like, find gradients better or just find better gradients.