Call for Papers
We invite papers and posters that describe innovative combinations of symbolic knowledge representation/reasoning approaches and learning, including new methodology, datasets, evaluations, surveys, reproduced results, and negative results. In addition, we also invite challenge proposals that specify challenges in the intersection of the two areas and/or outline a vision of how these could be tackled by the community at large.
Topics of Interest
The KR2ML workshop is intended to provide a forum for discussing innovative approaches and challenges in integrating KRR and ML approaches, and for exchanging ideas about how to move the area forward. We specifically invite challenge proposals to describe major problems in the integration of KRR and ML, and those that outline a vision of how these could be tackled by the community at large.
Topics of interest for all kinds of submissions include, but are not limited to:
- Successful integrations of ML and KRR-based techniques.
- Approaches that tackle problems in the integration of the two areas (e.g., evaluations, datasets).
- ML-based solutions to problems in KRR systems, or the other way round; for examples see below.
KRR ← ML
- Knowledge representation analysis via learned models
- The representation of symbolic knowledge by learning models
- Automatic learning of theories (e.g., action theories)
- Improved symbolic reasoning via ML methods
KRR → ML
- Symbolic priors (e.g., domain knowledge) to improve learning
- Symbolic representations for generating large amounts of training or simulation data
- Symbolic representations in learning models
- Integration of logic into learning models
- Extraction of symbolic knowledge from learning models (e.g., as explanations)
We have three types for submissions for the workshop:
- Full Papers (5-8 pages)
- Poster abstracts (3-4 pages)
- Challenge proposals (1-3 pages)
Submissions to KR2ML 2019 are limited to the above pages of content, and may contain any number of pages for references. Appendices can be included beyond the references in the same PDF file, but may not be considered by the reviewers.
The submission process is not strictly double-blind but, for works currently under double-blind review, we recommend the authors retain anonymity in their submitted work. The assigned reviewers will not be shared with the individual authors.
All submissions must be formatted with LaTeX using the NeurIPS paper format: https://nips.cc/Conferences/2019/PaperInformation/StyleFiles
The accepted papers will be published on the KR2ML homepage, and are non-archival. We welcome interesting articles published currently under review or those that are planned for publishing elsewhere (see also below topics of interest).
Submission Link: https://easychair.org/conferences/?conf=kr2ml2019
Please email any enquiries to email@example.com