Criteria of efficiency for set-valued classification
- Submitting institution
-
Royal Holloway and Bedford New College
- Unit of assessment
- 11 - Computer Science and Informatics
- Output identifier
- 29004167
- Type
- D - Journal article
- DOI
-
10.1007/s10472-017-9540-3
- Title of journal
- Annals of Mathematics and Artificial Intelligence
- Article number
- -
- First page
- 21
- Volume
- 81
- Issue
- -
- ISSN
- 1012-2443
- Open access status
- Compliant
- Month of publication
- March
- Year of publication
- 2017
- 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
-
4
- Research group(s)
-
-
- Citation count
- 1
- Proposed double-weighted
- No
- Reserve for an output with double weighting
- No
- Additional information
- Validity and efficiency have been major criteria to assess performance of conformal predictors. Of these two the efficiency required special attention since validity is achieved automatically. In the classification problem efficiency is measured by the size of the prediction set. However, this paper points out that the standard criteria of efficiency have serious disadvantages, for example, a non-uniqueness of choice function. The paper defines a class of criteria of efficiency, called “probabilistic”, that avoid this disadvantage and should be used in place of more standard ones.
- Author contribution statement
- -
- Non-English
- No
- English abstract
- -