Generalized low-rank models

[maxentile]

Looks like a really useful unifying perspective / approach:
http://arxiv.org/pdf/1410.0342v4.pdf
https://github.com/cehorn/glrm