What is it about?
We developed a technique that makes it possible for non-optimisation experts to include information about dependencies between data in their search space. Additionally, they can add ranges to these data to determine whether this dependency may only hold in a certain area of the search space. This information is automatically included in the optimisation algorithm to ensure the resulting solution to be valid in terms of the given dependencies.
Featured Image
Photo by Daniel Lerman on Unsplash
Why is it important?
Up until now, it was only possible to include dependencies in search space data for the entire search space. Additionally, this had to be encoded in the optimisation algorithm. Our work shows two major contributions: First, when real-world data has dependencies that are only valid for a certain part of the search space, but not for others - this can be included in the optimisation algorithm. Second, this can be done by the domain expert rather than the optimisation expert by using an easy language to specify this information. The optimisation algorithm automatically includes this information in its search.
Perspectives
To me, the simplicity for domain experts is really important. When a project ends, the synthesis between domain experts and optimisation experts usually ends as well. However, if a domain expert now discovers new information about the problem, he or she can easily modify the specification file in a language easily understandable for him or her, and the optimisation takes informatoin into account.
Christina Plump
Deutsches Forschungszentrum fur Kunstliche Intelligenz GmbH
Read the Original
This page is a summary of: Adapting mutation and recombination operators to range-aware relations in real-world application data, July 2022, ACM (Association for Computing Machinery),
DOI: 10.1145/3520304.3529066.
You can read the full text:
Contributors
The following have contributed to this page
