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

While deep learning has been very beneficial in data-rich settings, taskswith smaller training set often resort to pre-training or multitask learning toleverage data from other tasks. In this case, careful consideration is neededto select tasks and model parameterizations such that updates from theauxiliary tasks actually help the primary task. We seek to alleviate thisburden by formulating a model-agnostic framework that performs fine-grainedmanipulation of the auxiliary task gradients. We propose to decompose auxiliaryupdates into directions which help, damage or leave the primary task lossunchanged. This allows weighting the update directions differently depending ontheir impact on the problem of interest. We present a novel and efficientalgorithm for that purpose and show its advantage in practice. Our methodleverages efficient automatic differentiation procedures and randomizedsingular value decomposition for scalability. We show that our framework isgeneric and encompasses some prior work as particular cases. Our approachconsistently outperforms strong and widely used baselines when leveragingout-of-distribution data for Text and Image classification tasks.

Auxiliary Task Update Decomposition: The Good, The Bad and The Neutral