/CbAS

Code for the ICML 2019 paper 'Conditioning by adaptive sampling for robust design'

Primary LanguageJupyter Notebook

Conditioning by Adaptive Sampling for Robust Design

This repo contains the code for the paper:

D. H. Brookes, H. Park, and J. Listgarten. Conditioning by adaptive sampling for robust design. Proceedings of ICML, 2019.

The most important bits of code are in the files src/optimization_algs.py and notebooks/toy_conditioning.ipynb. In particular the function weighted_ml_opt in src/optimization_algs.py, with weights_type='cbas' runs the central CbAS method. Additionally, notebooks/toy_conditioning.ipynb, is a self-contained iPython notebook that runs the CbAS tests on the toy problem shown in Figure 1.