Multi Fidelity Entropy Search (MF-ES) synthetic example. This algorithm extends the original Entropy Search (ES) (code) to the multifidelity setting. The publication to which this code is given below

Alonso Marco, Felix Berkenkamp, Philipp Hennig, Angela P. Schoellig, Andreas Krause, Stefan Schaal, Sebastian Trimpe "Virtual vs. Real: Trading Off Simulations and Physical Experiments in Reinforcement Learning with Bayesian Optimization", IEEE International Conference on Robotics and Automation (ICRA), 2017, accepted.

The paper can be found here.

This project is a high-level user interface that needs from the Entropy Search library ESlib. to run. While ESlib contains all the computations of MF-ES, this project just allows the user to choose options of the global optimization problem, like maximum number of evaluations, GP kernel to be used, hyperparameters of the kernel, etc. For a full description on how to run the code, please refer to instructions.txt.

An alternative version that improves on efficiency of the original ES code and provides several plotting tools can be found here.

This package needs the Entropy Search (ES) library, which can be found here. It works under Matlab 2017 or higher.

For any questions, please, send an e-mail to:

[email protected]