Multi-modal Adaptive Mixture of Experts for Cold-start Recommendation Conference: Proceedings of the 34th ACM International Conference on Information and Knowledge Management; November 10–14, 2025; Seoul, Korea Proceedings of the 34th ACM International Conference on Information and Knowledge Managem
Van-Khang Nguyen, Duc-Hoang Pham, Huy-Son Nguyen, Cam-Van Thi Nguyen, Hoang-Quynh Le, Duc-Trong Le · 2025
Read Multi-modal Adaptive Mixture of Experts for Cold-start Recommendation Conference: Proceedings of the 34th ACM International Conference on Information and Knowledge Management; November 10–14, 2025; Seoul, Korea Proceedings of the 34th ACM International Conference on Information and Knowledge Managem by Van-Khang Nguyen, Duc-Hoang Pham, Huy-Son Nguyen, Cam-Van Thi Nguyen, Hoang-Quynh Le, Duc-Trong Le (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-2508-08042.txt.
Description
Nguyen and co-authors tackle the cold-start problem in recommendation: how to suggest a brand-new item that nobody has clicked yet. New items carry no usage history, but they often arrive with rich material — pictures, text, sometimes audio. Blending those signals naively, by gluing or averaging them, wastes their differences. The team's system, MAMEX, instead runs a mixture of experts. One expert network serves each kind of signal.
A learned gating mechanism decides, item by item, how much each expert's view should count. It favors whichever signal is most informative for that item and stays steady when one is weak or missing. On benchmark data sets, MAMEX beat the leading cold-start methods on accuracy and adaptability. The code is public. For platform teams, the paper offers a recommender design that listens to each kind of evidence on its own terms.
Catalog record
| CREATOR | Van-Khang Nguyen, Duc-Hoang Pham, Huy-Son Nguyen, Cam-Van Thi Nguyen, Hoang-Quynh Le, Duc-Trong Le |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 20 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2508-08042