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

Exploiting known semantic relationships between fine-grained tasks iscritical to the success of recent model agnostic approaches. These approachesoften rely on meta-optimization to make a model robust to systematic task ordomain shifts. However, in practice, the performance of these methods cansuffer, when there are no coherent semantic relationships between the tasks (ordomains). We present Invenio, a structured meta-learning algorithm to infersemantic similarities between a given set of tasks and to provide insights intothe complexity of transferring knowledge between different tasks. In contrastto existing techniques such as Task2Vec and Taskonomy, which measuresimilarities between pre-trained models, our approach employs a novelself-supervised learning strategy to discover these relationships in thetraining loop and at the same time utilizes them to update task-specific modelsin the meta-update step. Using challenging task and domain databases, underfew-shot learning settings, we show that Invenio can discover intricatedependencies between tasks or domains, and can provide significant gains overexisting approaches in terms of generalization performance. The learnedsemantic structure between tasks/domains from Invenio is interpretable and canbe used to construct meaningful priors for tasks or domains.

Invenio: Discovering Hidden Relationships Between Tasks/Domains Using Structured Meta Learning