Kernel combination via debiased object correspondence analysis
- Submitting institution
-
Middlesex University
- Unit of assessment
- 11 - Computer Science and Informatics
- Output identifier
- 1221
- Type
- D - Journal article
- DOI
-
10.1016/j.inffus.2015.02.002
- Title of journal
- Information Fusion
- Article number
- -
- First page
- 228
- Volume
- 27
- Issue
- -
- ISSN
- 1566-2535
- Open access status
- Out of scope for open access requirements
- Month of publication
- March
- Year of publication
- 2015
- URL
-
http://eprints.mdx.ac.uk/15310/
- 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)
-
-
- Citation count
- 2
- Proposed double-weighted
- No
- Reserve for an output with double weighting
- No
- Additional information
- Kernel methods provide a very powerful and universal way to combine data of completely different types for machine-learning. However, it is only possible to combine kernels when there are matching objects in each domain, which can be very hard to achieve in practice due to propriety databases or missing information. The significance of this paper lies on its proposal of a completely novel technique for generating kernel objects common to all domains in a maximally unbiased manner. Tests on varied real-world data show that the approach achieves substantially better performance than the alternatives of data-pruning or decision-combination.
- Author contribution statement
- -
- Non-English
- No
- English abstract
- -