They are humble. What they do not say:

> used by its author to win the Higgs Boson Machine Learning Challenge.

The author also won Google AI Challenge in 2010. Using Common Lisp, but without this library, as no machine learning was needed. https://en.m.wikipedia.org/wiki/AI_Challenge There is a talk about his approach in that competition: https://www.youtube.com/watch?v=7sgERtZkycU

Apparently, some other competitions are also won: https://www.kaggle.com/melisgl

This is a really impressively put together piece of software and a great showcase for the MGL-PAX self-documentation framework. I'm getting a lot of inspiration from this stack lately.

Yeah I was going to comment that I was kind of astounded by the quality of the documentation. And though I haven't looked, that gives me high hopes for the code itself too. Usually well documented code tends to be well written code.

I find myself inspired as well. We should all aspire to document our code this thoroughly.

I'm a newcomer to this subject and I'm finding it very refreshing to have such a well-taped path ahead of me.

Please note that despite current lack of MGL-PAX (https://github.com/melisgl/mgl-pax) documentation, Boltzman Machines and Gaussian Processes are implemented and documented in the source, which is very well and cleanly written

https://github.com/melisgl/mgl/blob/master/src/boltzmann-mac...

https://github.com/melisgl/mgl/blob/master/src/gaussian-proc...