English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Recent studies on few-shot classification using transfer learning posechallenges to the effectiveness and efficiency of episodic meta-learningalgorithms. Transfer learning approaches are a natural alternative, but theyare restricted to few-shot classification. Moreover, little attention has beenon the development of probabilistic models with well-calibrated uncertaintyfrom few-shot samples, except for some Bayesian episodic learning algorithms.To tackle the aforementioned issues, we propose a new transfer learning methodto obtain accurate and reliable models for few-shot regression andclassification. The resulting method does not require episodic meta-learningand is called meta-free representation learning (MFRL). MFRL first findslow-rank representation generalizing well on meta-test tasks. Given the learnedrepresentation, probabilistic linear models are fine-tuned with few-shotsamples to obtain models with well-calibrated uncertainty. The proposed methodnot only achieves the highest accuracy on a wide range of few-shot learningbenchmark datasets but also correctly quantifies the prediction uncertainty. Inaddition, weight averaging and temperature scaling are effective in improvingthe accuracy and reliability of few-shot learning in existing meta-learningalgorithms with a wide range of learning paradigms and model architectures.

Meta-free few-shot learning via representation learning with weight averaging