[maemre]

Stochastic optimization? STOKE [1] from Stanford is trying to that if I understood it correctly.

[1]: https://github.com/StanfordPL/stoke