MetaBO

Dear reviewers,

this is the source code accompanying the ICLR 2020 Submission: "Meta-Learning Acquisition Functions for Transfer Learning in Bayesian Optimization".

Installation

We kindly ask you to clone this repository and run

conda env create -f environment.yml

to create a new conda environment named "metabo" with all python packages required to run the experiments.

Scripts to reproduce the results presented in the paper. These scripts are named evaluate_metabo_<experiment_name>.py. They load pre-trained network weights stored in /metabo/iclr2020/<experiment_name> to reproduce the results without the need of re-training neural acquisition functions.
Scripts to re-train the aforementioned neural acquisition functions. These scripts are named train_metabo_<experiment_name>.py.

Copyright holder of the paper "Meta-Learning Acquisition Functions for Transfer Learning in Bayesian Optimization".

Submitted to ICLR 2020 for review.

Name		Name	Last commit message	Last commit date
Latest commit History 2 Commits
metabo		metabo
.gitignore		.gitignore
README.md		README.md
environment.yml		environment.yml
evaluate_metabo_adaboost.py		evaluate_metabo_adaboost.py
evaluate_metabo_branin.py		evaluate_metabo_branin.py
evaluate_metabo_furuta.py		evaluate_metabo_furuta.py
evaluate_metabo_gprice.py		evaluate_metabo_gprice.py
evaluate_metabo_gps.py		evaluate_metabo_gps.py
evaluate_metabo_hm3.py		evaluate_metabo_hm3.py
evaluate_metabo_svm.py		evaluate_metabo_svm.py
train_metabo_adaboost.py		train_metabo_adaboost.py
train_metabo_branin.py		train_metabo_branin.py
train_metabo_furuta.py		train_metabo_furuta.py
train_metabo_gprice.py		train_metabo_gprice.py
train_metabo_gps.py		train_metabo_gps.py
train_metabo_hm3.py		train_metabo_hm3.py
train_metabo_svm.py		train_metabo_svm.py