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

In Location-based Social Networks, Point-of-Interest (POI) recommendationhelps users discover interesting places. There is a trend to move from thecloud-based model to on-device recommendations for privacy protection andreduced server reliance. Due to the scarcity of local user-item interactions onindividual devices, solely relying on local instances is not adequate.Collaborative Learning (CL) emerges to promote model sharing among users, wherereference data is an intermediary that allows users to exchange their softdecisions without directly sharing their private data or parameters, ensuringprivacy and benefiting from collaboration. However, existing CL-basedrecommendations typically use a single reference for all users. Reference datavaluable for one user might be harmful to another, given diverse userpreferences. Users may not offer meaningful soft decisions on items outsidetheir interest scope. Consequently, using the same reference data for allcollaborations can impede knowledge exchange and lead to sub-optimalperformance. To address this gap, we introduce the Decentralized CollaborativeLearning with Adaptive Reference Data (DARD) framework, which crafts adaptivereference data for effective user collaboration. It first generates adesensitized public reference data pool with transformation and probabilitydata generation methods. For each user, the selection of adaptive referencedata is executed in parallel by training loss tracking and influence function.Local models are trained with individual private data and collaboratively withthe geographical and semantic neighbors. During the collaboration between twousers, they exchange soft decisions based on a combined set of their adaptivereference data. Our evaluations across two real-world datasets highlight DARD'ssuperiority in recommendation performance and addressing the scarcity ofavailable reference data.

Decentralized Collaborative Learning with Adaptive Reference Data for On-Device POI Recommendation