XLearn : learning activity labels across heterogeneous datasets
- Submitting institution
-
University of St Andrews
- Unit of assessment
- 11 - Computer Science and Informatics
- Output identifier
- 266336147
- Type
- D - Journal article
- DOI
-
10.1145/3368272
- Title of journal
- ACM Transactions on Intelligent Systems and Technology
- Article number
- 17
- First page
- -
- Volume
- 11
- Issue
- 2
- ISSN
- 2157-6904
- Open access status
- Compliant
- Month of publication
- January
- 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
-
2
- Research group(s)
-
A - Artificial Intelligence
- Citation count
- 1
- Proposed double-weighted
- No
- Reserve for an output with double weighting
- No
- Additional information
- Machine learning is not usually compositional: you learn a classifier from a single large dataset, which is a barrier to entry and a problem both for scaling and for transferring classifiers to different environments. This work demonstrates that human-activity classifiers can be learned separately and combined, so that a more general classifier can be built incrementally from smaller ones. This is a major step towards a compositional approach to machine learning.
- Author contribution statement
- -
- Non-English
- No
- English abstract
- -
