dc.citation.conferencePlace |
IT |
- |
dc.citation.conferencePlace |
Venice Convention CenterVenice |
- |
dc.citation.endPage |
3581 |
- |
dc.citation.startPage |
3573 |
- |
dc.citation.title |
IEEE International Conference on Computer Vision |
- |
dc.contributor.author |
Kim, Kwang In |
- |
dc.contributor.author |
Tompkin, James |
- |
dc.contributor.author |
Richardt, Christian |
- |
dc.date.accessioned |
2023-12-19T18:08:05Z |
- |
dc.date.available |
2023-12-19T18:08:05Z |
- |
dc.date.created |
2019-02-28 |
- |
dc.date.issued |
2017-10-22 |
- |
dc.description.abstract |
We present an algorithm for test-time combination of a set of reference predictors with unknown parametric forms. Existing multi-task and transfer learning algorithms focus on training-time transfer and combination, where the parametric forms of predictors are known and shared. However, when the parametric form of a predictor is unknown, e.g., for a human predictor or a predictor in a precompiled library, existing algorithms are not applicable. Instead, we empirically evaluate predictors on sampled data points to measure distances between different predictors. This embeds the set of reference predictors into a Riemannian manifold, upon which we perform manifold denoising to obtain the refined predictor. This allows our approach to make no assumptions about the underlying predictor forms. Our test-time combination algorithm equals or outperforms existing multi-task and transfer learning algorithms on challenging real-world datasets, without introducing specific model assumptions. © 2017 IEEE. |
- |
dc.identifier.bibliographicCitation |
IEEE International Conference on Computer Vision, pp.3573 - 3581 |
- |
dc.identifier.doi |
10.1109/ICCV.2017.384 |
- |
dc.identifier.issn |
1550-5499 |
- |
dc.identifier.scopusid |
2-s2.0-85041921195 |
- |
dc.identifier.uri |
https://scholarworks.unist.ac.kr/handle/201301/32671 |
- |
dc.identifier.url |
https://ieeexplore.ieee.org/document/8237646 |
- |
dc.language |
영어 |
- |
dc.publisher |
IEEE |
- |
dc.title |
Predictor Combination at Test Time |
- |
dc.type |
Conference Paper |
- |
dc.date.conferenceDate |
2017-10-22 |
- |