Dynamic Causal Bayesian Optimization

This is a Python implementation of Dynamic Causal Bayesian Optimization as presented at NeurIPS 2021.

Abstract

This paper studies the problem of performing a sequence of optimal interventions in a causal dynamical system where both the target variable of interest and the inputs evolve over time. This problem arises in a variety of domains e.g. system biology and operational research. Dynamic Causal Bayesian Optimization (DCBO) brings together ideas from sequential decision making, causal inference and Gaussian process (GP) emulation. DCBO is useful in scenarios where all causal effects in a graph are changing over time. At every time step DCBO identifies a local optimal intervention by integrating both observational and past interventional data collected from the system. We give theoretical results detailing how one can transfer interventional information across time steps and define a dynamic causal GP model which can be used to quantify uncertainty and find optimal interventions in practice. We demonstrate how DCBO identifies optimal interventions faster than competing approaches in multiple settings and applications.

Authors

Virginia Aglietti, Neil Dhir, Javier Gonzalez and Theodoros Damoulas

Demonstration

Installation

pip install -e .

Tutorials

• stat_scm.ipynb walks through the basic steps required to run all methods (BO, ABO, CBO and DCBO) on the example SCM shown in figure 1 of the paper (shown above in panel (a)).
• nonstat_scm.ipynb demonstrates how to run the methods on the SCM with the DAG shown in figure 3(c) of the paper, but over multiple replicates, to allow for confidence bounds over the results.
• ind_scm.ipynb demonstrates how to run the methods on the SCM with the DAG shown in figure 3(b) of the paper, but over multiple replicates, to allow for confidence bounds over the results. In this notebook we show a more conventional experiment. Importantly we demonstrate it for the SCM in which DCBO was not the most successful method.
• A more extensive cyber-security example can be found in this repository which was introduced in this paper - here DCBO acts as a blue agent, taking optimal actions to defend a computer network.

Citation

Please cite the NeurIPS paper if you use DCBO in your work:

@inproceedings{DCBO,
 author = {Aglietti, Virginia and Dhir, Neil and Gonz\'{a}lez, Javier and Damoulas, Theodoros},
 booktitle = {Advances in Neural Information Processing Systems},
 title = {Dynamic Causal Bayesian Optimization},
 volume = {35},
 year = {2021}
}

License

This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation; either version 3 of the License, or (at your option) any later version.

This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.

You should have received a copy of the GNU General Public License along with this program. If not, see http://www.gnu.org/licenses/.

Disclaimer

This is research code. Written by researchers, for researchers. We are not professional software developers and are thus not equipped to make a (more) robust implementation. Consequently, things may break. If they do please don't get angry. Instead, if you would like to contribute, find things that are broken or have any suggestions for this work, you can contact us at ndhir@turing.ac.uk or open an issue on this repository.