This is code to accompany the paper: From Complexity to Simplicity: Adaptive ES-Active Subspaces for Blackbox Optimization implementing one version of the ASEBO algorithm. Here is the link to the paper:
It is not the implementation used for the experiments in the paper, which is proprietary, but is a re-implementation which has been tested. The benefit of this is that we use numpy and a single machine, so it should run locally on a laptop.
We hope you find this useful - if you do, please cite:
@incollection{asebo,
title = {From Complexity to Simplicity: Adaptive ES-Active Subspaces for Blackbox Optimization},
author = {Choromanski, Krzysztof M and Pacchiano, Aldo and Parker-Holder, Jack and Tang, Yunhao and Sindhwani, Vikas},
booktitle = {Advances in Neural Information Processing Systems 32},
year = {2019}
}