search

Yelp / MOE

A global, black box optimization engine for real world metric optimization.
Other
1.3k stars 139 forks source link

MOE vs PyKriging #467

Open tomaso909 opened 7 years ago

tomaso909 commented 7 years ago

I want to explore the effect that different variables have on a black-box function and eventually find its global minimum. The function is computationally expensive to evaluate so I'd like to run the minimum number of computer simulations.

In the past I have used a python implementation of the Kriging algorithm (see http://pykriging.com/). I am wondering if there is any advantage on using MOE in my case. Since both use gaussian processes, I think the results should be the same, but I'd like to know your opinions. Thanks.

themrmax commented 7 years ago

MOE implements the kriging algorithm (gp_next_points_kriging) and also implements some other algoritms (EPI and constant liar). So you should use MOE if you want to try these out too. Also you may or may not prefer MOE's API over the other library you mention.