Meta-learning Tree-structured Parzen Estimator

This package was used for the experiments of the paper Speeding up Multi-objective Hyperparameter Optimization by Task Similarity-Based Meta-Learning for the Tree-structured Parzen Estimator.

Note

Please check examples for the running example.

Setup

This package requires python 3.8 or later version. You can install the dependencies by:

$ conda create -n meta-learn-tpe python==3.8
$ pip install -r requirements.txt

# Create a directory for tabular datasets
$ mkdir ~/tabular_benchmarks
$ cd ~/tabular_benchmarks

# The download of HPOLib
$ cd ~/tabular_benchmarks
$ wget http://ml4aad.org/wp-content/uploads/2019/01/fcnet_tabular_benchmarks.tar.gz
$ tar xf fcnet_tabular_benchmarks.tar.gz
$ mv fcnet_tabular_benchmarks hpolib

Running example

The data obtained in the experiments are reproduced by the following command:

$ ./run_experiment.sh -s 0 -d 19

Citations

For the citation, use the following format:

@article{watanabe2023speeding,
  title={Speeding up Multi-objective Hyperparameter Optimization by Task Similarity-Based Meta-Learning for the Tree-structured {P}arzen Estimator},
  author={S. Watanabe and N. Awad and M. Onishi and F. Hutter},
  journal={International Joint Conference on Artificial Intelligence},
  year={2023}
}