Efficient Multimodal Streaming Recommendation via Expandable Side Mixture-of-Experts DOI: XXXXXXX.XXXXXXX Conference: Make sure to enter the correct conference title from your rights confirmation email; November 10–14, 2025; Seoul, Korea ISBN: 978-1-4503-XXXX-X/18/06 CCS: Information systems Recomme

Yunke Qu, Liang Qu, Tong Chen, Quoc Viet Hung Nguyen, Hongzhi Yin · 2025

Read Efficient Multimodal Streaming Recommendation via Expandable Side Mixture-of-Experts DOI: XXXXXXX.XXXXXXX Conference: Make sure to enter the correct conference title from your rights confirmation email; November 10–14, 2025; Seoul, Korea ISBN: 978-1-4503-XXXX-X/18/06 CCS: Information systems Recomme by Yunke Qu, Liang Qu, Tong Chen, Quoc Viet Hung Nguyen, Hongzhi Yin (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2508-05993.txt.

Description

Yunke Qu and co-authors keep streaming recommenders fresh without breaking the bank in a 2025 paper. Apps that suggest items must track tastes that shift daily, while new items arrive with no history. Rich image and text features help, but tuning the big encoders that make them is costly.

Their frame adds small side modules to frozen encoders. The modules hold banks of expert nets that grow as tastes change. A router blends expert and base output per item. Idle experts are pruned to keep the model lean. New patterns are learned in the side banks, so old taste knowledge is not overwritten. The design targets live services where both memory and freshness count.

Catalog record

CREATORYunke Qu, Liang Qu, Tong Chen, Quoc Viet Hung Nguyen, Hongzhi Yin
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE55 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2508-05993

Browse the full library · About the Tower of Babel