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

Modern platforms leverage randomized experiments to make informed decisionsfrom a given set of items (``treatments''). As a particularly challengingscenario, these items may (i) arrive in high volume, with thousands of newitems being released per hour, and (ii) have short lifetime, say, due to theitem's transient nature or underlying non-stationarity that impels the platformto perceive the same item as distinct copies over time. Thus motivated, westudy a Bayesian multiple-play bandit problem that encapsulates the keyfeatures of the multivariate testing (or ``multi-A/B testing'') problem with ahigh volume of short-lived arms. In each round, a set of $k$ arms arrive, eachavailable for $w$ rounds. Without knowing the mean reward for each arm, thelearner selects a multiset of $n$ arms and immediately observes their realizedrewards. We aim to minimize the loss due to not knowing the mean rewards,averaged over instances generated from a given prior distribution. We show thatwhen $k = O(n^\rho)$ for some constant $\rho>0$, our proposed policy has$\tilde O(n^{-\min \{\rho, \frac 12 (1+\frac 1w)^{-1}\}})$ loss on asufficiently large class of prior distributions. We complement this result byshowing that every policy suffers $\Omega (n^{-\min \{\rho, \frac 12\}})$ losson the same class of distributions. We further validate the effectiveness ofour policy through a large-scale field experiment on {\em Glance}, acontent-card-serving platform that faces exactly the above challenge. A simplevariant of our policy outperforms the platform's current recommender by 4.32\%in total duration and 7.48\% in total number of click-throughs.

Short-lived High-volume Multi-A(rmed)/B(andits) Testing