Trepan reloaded: A knowledge-driven approach to explaining black-box models
- Submitting institution
-
City, University of London
- Unit of assessment
- 11 - Computer Science and Informatics
- Output identifier
- 1248
- Type
- E - Conference contribution
- DOI
-
10.3233/FAIA200378
- Title of conference / published proceedings
- ECAI 2020: 24th European Conference on Artificial Intelligence (ECAI 2020)
- First page
- 2457
- Volume
- 325
- Issue
- -
- ISSN
- 0922-6389
- Open access status
- Compliant
- Month of publication
- August
- Year of publication
- 2020
- URL
-
-
- Supplementary information
-
-
- Request cross-referral to
- -
- Output has been delayed by COVID-19
- No
- COVID-19 affected output statement
- -
- Forensic science
- No
- Criminology
- No
- Interdisciplinary
- No
- Number of additional authors
-
3
- Research group(s)
-
-
- Citation count
- -
- Proposed double-weighted
- No
- Reserve for an output with double weighting
- No
- Additional information
- Output recognised with Distinguished Paper Award at Europe's premier AI conference, ECAI 2020. The reported work is significant because it is the first approach to combining explanation using the Trepan Algorithm with prior knowledge in the form of an ontology. The novel method presented enables more effective explanations of black-box models, as clearly shown in the user study. This is the result of a collaboration with Telefonica Alpha where the practical need for this research arose.
- Author contribution statement
- -
- Non-English
- No
- English abstract
- -