Neural network vs computing by eye

Today I learned that people can by eye estimate the rank of the matrix!
(remember, rank is the number of linearly independent rows or columns)

Like, look at this and say that rank -

How to resist and not try to train to do this neural network, thought ukurki students from Carnegie Mellon?

That is, take a matrix, translate it into such a picture and give it to the CNN input so that it predicts the rank or whether the matrix is ​​degenerate.

Look at the bald. Here it is, the future of algorithm optimization.

The article here and in it are great and the syllable, and the questions raised, and the list of authors, and everything.

For example, further they apply the same approach to matrix multiplication and finding the inverse:

We then use this data.
stochastic gradient descent on a mean square error
(MSE) loss for 100 epochs. Some qualitative predictions
on unseen data are shown in Figures 7 and 8.
Found
by our network architecture, but the inversion task
proved much more challenging, as shown by the
higher MSE values. We note that this is analogous to
humans taking Linear Algebra 101 .

Rzhu Nimaga.
For sweets - slides .