Precision-Recall-Gain Curves : PR Analysis Done Right
- Submitting institution
-
University of Bristol
- Unit of assessment
- 11 - Computer Science and Informatics
- Output identifier
- 95745371
- Type
- E - Conference contribution
- DOI
-
-
- Title of conference / published proceedings
- Advances in Neural Information Processing Systems 28
- First page
- 838
- Volume
- -
- Issue
- -
- ISSN
- -
- Open access status
- -
- Month of publication
- December
- Year of publication
- 2015
- 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
-
1
- Research group(s)
-
A - Artificial Intelligence and Autonomy
- Citation count
- 4
- Proposed double-weighted
- No
- Reserve for an output with double weighting
- No
- Additional information
- This paper identifies and corrects fundamental flaws of precision-recall curves, widely used in machine learning, information retrieval and computer vision. These flaws often result in sub-optimal model selection; the novel PRG curves avoid this, with significant and growing methodological impact as evident from citation analysis. The work is one of the main outputs of the REFRAME project funded by CHIST-ERA (https://www.chistera.eu/projects/reframe, 2012-15, €950k for Bristol-Valencia-Strasbourg). Selected for a plenary oral 'spotlight' presentation at NIPS (17% acceptance rate). The work has led to a follow-on project funded by the Turing Institute (https://www.turing.ac.uk/research/research-projects/measurement-theory-data-science-and-ai).
- Author contribution statement
- -
- Non-English
- No
- English abstract
- -