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 Tools annual

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 Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

Extreme multi-label classification (XMC) aims to learn a model that can tagdata points with a subset of relevant labels from an extremely large label set.Real world e-commerce applications like personalized recommendations andproduct advertising can be formulated as XMC problems, where the objective isto predict for a user a small subset of items from a catalog of several millionproducts. For such applications, a common approach is to organize these labelsinto a tree, enabling training and inference times that are logarithmic in thenumber of labels. While training a model once a label tree is available is wellstudied, designing the structure of the tree is a difficult task that is notyet well understood, and can dramatically impact both model latency andstatistical performance. Existing approaches to tree construction fall at anextreme point, either optimizing exclusively for statistical performance, orfor latency. We propose an efficient information theory inspired algorithm toconstruct intermediary operating points that trade off between the benefits ofboth. Our algorithm enables interpolation between these objectives, which wasnot previously possible. We corroborate our theoretical analysis with numericalresults, showing that on the Wiki-500K benchmark dataset our method can reducea proxy for expected latency by up to 28% while maintaining the same accuracyas Parabel. On several datasets derived from e-commerce customer logs, ourmodified label tree is able to improve this expected latency metric by up to20% while maintaining the same accuracy. Finally, we discuss challenges inrealizing these latency improvements in deployed models.

Enabling Efficiency-Precision Trade-offs for Label Trees in Extreme Classification