steven-adriaensen / is4ape

Library implementing Importance Sampling for Algorithm Performance Estimation

Importance Sampling for Algorithm Performance Estimation (IS4APE)

This project contains the source code for the research article:
"An Importance Sampling Approach to the Estimation of Algorithm Performance in Automated Algorithm Design" (Adriaensen et al, 2017)

Feel free to use IS4APE in your own research. If you do, we kindly ask you to acknowledge this, by citing the following article:

Adriaensen, Steven, Filip Moons, and Ann Nowé. "An Importance Sampling Approach to the Estimation of Algorithm Performance in Automated Algorithm Design." International Conference on Learning and Intelligent Optimization. Springer, Cham, 2017.

@inproceedings{adriaensen2017importance,
title={An Importance Sampling Approach to the Estimation of Algorithm Performance in Automated Algorithm Design},
author={Adriaensen, Steven and Moons, Filip and Now{'e}, Ann},
booktitle={International Conference on Learning and Intelligent Optimization},
pages={3--17},
year={2017},
organization={Springer}
}

Content:

See the documentation in these files for more details.

Links:

https://link.springer.com/chapter/10.1007/978-3-319-69404-7_1

UPDATE: What's new:

• implementations of two additional experimental scenarios:
  • the design of a (static) sorting portfolio (to use, please extract sort_data.zip into the sort_data directory)
  • the design of a dynamic metaheuristic scheduler (to use, please extract hh_data.zip into the hh_data directory)
• improvements to the PoC and IS estimates.

(the original code used in (Adriaensen et al, 2017) can still be found in directory "lion2017")