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

Pre-trained recommendation models (PRMs) have attracted widespread attentionrecently. However, their totally different model structure, huge model size andcomputation cost hinder their application in practical recommender systems.Hence, it is highly essential to explore how to practically utilize PRMs inreal-world recommendations. In this paper, we propose a novel joint knowledgedistillation from different pre-trained recommendation models named PRM-KD forrecommendation, which takes full advantages of diverse PRMs as teacher modelsfor enhancing student models efficiently. Specifically, PRM-KD jointly distillsdiverse informative knowledge from multiple representative PRMs such asUniSRec, Recformer, and UniM^2Rec. The knowledge from the above PRMs are thensmartly integrated into the student recommendation model considering theirconfidence and consistency. We further verify the universality of PRM-KD withvarious types of student models, including sequential recommendation, featureinteraction, and graph-based models. Extensive experiments on five real-worlddatasets demonstrate the effectiveness and efficacy of PRM-KD, which could beviewed as an economical shortcut in practically and conveniently making fulluse of different PRMs in online systems.

Distillation is All You Need for Practically Using Different Pre-trained Recommendation Models