Code for "Robust Bayesian Satisficing", (NeurIPS 2023)

Setup

conda env create --name rbs --file environment.yml

Reproduction

For each of the simulations described in our paper, we have provided a corresponding Jupyter Notebook. Below is a list of the notebooks and a brief description of each:

• synthetic_exp.ipynb: This notebook demonstrates the proof of concept experiment.

• insulin_dosage_exp.ipynb: This notebook demonstrates the insulin dosage experiment with both setup scenarios.

• sensitivity_tau_vs_eps_exp.ipynb: This notebook provides the sensitivity analysis of $\tau$ for RoBOS and $r$ for DRBO.

Note: Please make sure to follow the comments within each notebook to ensure smooth execution.

You can cite RBS as below:

@inproceedings{
  saday2023rbs,
  title={Robust Bayesian Satisficing},
  author={Saday, Artun and Yıldırım, Y. Cahit and Tekin, Cem},
  booktitle={Advances in Neural Information Processing Systems 37},
  year={2023},
  url={https://arxiv.org/abs/2308.08291}
}