dc.citation.conferencePlace |
CN |
- |
dc.citation.conferencePlace |
Montreal |
- |
dc.citation.title |
Workshop on Meta-Learning (MetaLearn 2018) |
- |
dc.contributor.author |
Yoon, Sung Whan |
- |
dc.contributor.author |
Seo, Jun |
- |
dc.contributor.author |
Moon, Jaekyun |
- |
dc.date.accessioned |
2024-02-01T00:41:15Z |
- |
dc.date.available |
2024-02-01T00:41:15Z |
- |
dc.date.created |
2020-03-26 |
- |
dc.date.issued |
2018-12-08 |
- |
dc.description.abstract |
We propose a meta-learning algorithm utilizing a linear transformer that carries out null-space projection of neural network outputs. The main idea is to construct an alternative classification space such that the error signals during few-shot learning are quickly zero-forced on that space so that reliable classification on low data is possible. The final decision on a query is obtained utilizing a null-spaceprojected distance measure between the network output and reference vectors, both of which have been trained in the initial learning phase. Among the known methods with a given model size, our meta-learner achieves the best or near-best image classification accuracies with Omniglot and miniImageNet datasets. |
- |
dc.identifier.bibliographicCitation |
Workshop on Meta-Learning (MetaLearn 2018) |
- |
dc.identifier.uri |
https://scholarworks.unist.ac.kr/handle/201301/80302 |
- |
dc.identifier.url |
https://arxiv.org/abs/1806.01010 |
- |
dc.language |
영어 |
- |
dc.publisher |
NeurIPS |
- |
dc.title |
Meta-Learner with Linear Nulling |
- |
dc.type |
Conference Paper |
- |
dc.date.conferenceDate |
2018-12-08 |
- |