Sparkle
Sparkle is a problem solving and competition platform providing easy access to automatic algorithm selection and automatic algorithm configuration for experts and non-experts. The Sparkle platform is under active development, and so far includes the following features:
- Automated algorithm selection
- Automated algorithm configuration
- Ablation analysis
- Detailed result reports
For additional information or questions, please contact us at: sparkle@aim.rwth-aachen.de
Download and Source
The latest stable version of Sparkle can be downloaded here, and installation instructions are available here. Source code and documentation are also available from the git repository on: bitbucket.
Documentation
Documentation for Sparkle is available as PDF on the git repository and also on: readthedocs.
Challenges
-
Sparkle Planning Challenge 2019.Organisers: Chuan Luo, Mauro Vallati and Holger Hoos.[ website ]
-
Sparkle SAT Challenge 2018.Organisers: Chuan Luo and Holger Hoos.[ website ]
Material
- Koen van der Blom, Holger Hoos, Chuan Luo, Jeroen Rook.
Sparkle: Towards Accessible Meta-Algorithmics for Improving the State of the Art in Solving Challenging Problems.IEEE Transactions on Evolutionary Computation, 26(6):1351-1364, 2022.
- Koen van der Blom, Jeroen Rook.
Automated Algorithm Configuration and Selection with Sparkle.Tutorial, PPSN Conference, September 2022.[ PDF ]
- Koen van der Blom, Holger Hoos, Chuan Luo, Jeroen Rook.
Sparkle: Accessible Meta-Algorithmics for Improving the State of the Art in Solving Challenging Problems.Poster, TAILOR Conference #1, September 2021.[ PDF ]
- Holger Hoos.
Sparkle: A PbO-based Multi-agent Problem-solving Platform.Technical Report TR-2015-??, Department of Computer Science, University of British Columbia, December 2015.[ PDF ]
Contact
Contributors
See Bitbucket.