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 Tools annual

Global: Zendy Free Plan

Global: Zendy Tools monthly

Global: Zendy Plus monthly

Although existing Learning-to-Defer (L2D) frameworks support multipleexperts, they allocate each query to a single expert, limiting their ability toleverage collective expertise in complex decision-making scenarios. To addressthis, we introduce the first framework for Top-$k$ Learning-to-Defer, enablingsystems to defer each query to the $k$ most cost-effective experts. Ourformulation strictly generalizes classical two-stage L2D by supportingmulti-expert deferral-a capability absent in prior work. We further proposeTop-$k(x)$ Learning-to-Defer, an adaptive extension that learns the optimalnumber of experts per query based on input complexity, expert quality, andconsultation cost. We introduce a novel surrogate loss that isBayes-consistent, $(\mathcal{R}, \mathcal{G})$-consistent, and independent ofthe cardinality parameter $k$, enabling efficient reuse across different valuesof $k$. We show that classical model cascades arise as a special case of ourmethod, situating our framework as a strict generalization of both selectivedeferral and cascaded inference. Experiments on classification and regressiondemonstrate that Top-$k$ and Top-$k(x)$ yield improved accuracy--costtrade-offs, establishing a new direction for multi-expert deferral inLearning-to-Defer.

Why Ask One When You Can Ask $k$? Two-Stage Learning-to-Defer to the Top-$k$ Experts