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

Multi-armed bandit (MAB) algorithms are efficient approaches to reduce theopportunity cost of online experimentation and are used by companies to findthe best product from periodically refreshed product catalogs. However, thesealgorithms face the so-called cold-start at the onset of the experiment due toa lack of knowledge of customer preferences for new products, requiring aninitial data collection phase known as the burn-in period. During this period,MAB algorithms operate like randomized experiments, incurring large burn-incosts which scale with the large number of products. We attempt to reduce theburn-in by identifying that many products can be cast into two-sided products,and then naturally model the rewards of the products with a matrix, whose rowsand columns represent the two sides respectively. Next, we design two-phasebandit algorithms that first use subsampling and low-rank matrix estimation toobtain a substantially smaller targeted set of products and then apply a UCBprocedure on the target products to find the best one. We theoretically showthat the proposed algorithms lower costs and expedite the experiment in caseswhen there is limited experimentation time along with a large product set. Ouranalysis also reveals three regimes of long, short, and ultra-short horizonexperiments, depending on dimensions of the matrix. Empirical evidence fromboth synthetic data and a real-world dataset on music streaming servicesvalidates this superior performance.

Speed Up the Cold-Start Learning in Two-Sided Bandits with Many Arms