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

Transfer learning (TL) leverages previously obtained knowledge to learn newtasks efficiently and has been used to train deep learning (DL) models withlimited amount of data. When TL is applied to DL, pretrained (teacher) modelsare fine-tuned to build domain specific (student) models. This fine-tuningrelies on the fact that DL model can be decomposed to classifiers and featureextractors, and a line of studies showed that the same feature extractors canbe used to train classifiers on multiple tasks. Furthermore, recent studiesproposed multiple algorithms that can fine-tune teacher models' featureextractors to train student models more efficiently. We note that regardless ofthe fine-tuning of feature extractors, the classifiers of student models aretrained with final outputs of feature extractors (i.e., the outputs ofpenultimate layers). However, a recent study suggested that feature maps inResNets across layers could be functionally equivalent, raising the possibilitythat feature maps inside the feature extractors can also be used to trainstudent models' classifiers. Inspired by this study, we tested if feature mapsin the hidden layers of the teacher models can be used to improve the studentmodels' accuracy (i.e., TL's efficiency). Specifically, we developed 'adaptivetransfer learning (ATL)', which can choose an optimal set of feature maps forTL, and tested it in the few-shot learning setting. Our empirical evaluationssuggest that ATL can help DL models learn more efficiently, especially whenavailable examples are limited.

Adaptive Transfer Learning: a simple but effective transfer learning