arXiv：cs.LG· Xuesi Wang, Yangbin Shi, Xiaolin Zheng·· 4 小时前AI 评分36
生成式推荐中"遗漏目标"训练：将监督信号与概率竞争分开
Training with Missed Targets in Generative Recommendation: Separating Supervision from Probability Competition
AI 导读
生成式推荐器会在重排序前遗漏已观测目标，把遗漏目标追加到重排序训练列表会同时改变检索目标权重、引入追加目标监督并让两组争夺概率。研究用三个匹配损失将监督与组间竞争分离，实验显示这种竞争会损害已返回物品的排序：在四个预设的 Amazon Video Games 对比中移除竞争使 FT-NDCG 提升 7.8–22.2%。因此候选补全应针对每个生成器单独评估，而非自动套用。
正文
Abstract:Generative recommenders return a limited candidate set and may omit observed targets before reranking. A training strategy appends these missed targets to reranker training lists, although inference still ranks only original candidates. This operation simultaneously changes retrieved-target weight, adds supervision over appended targets, and makes the two groups compete for probability. An append/no-append comparison therefore cannot explain changes in returned-item rankings. We construct three matched losses that hold retrieved-target weight fixed while introducing appended-target supervision and group competition separately. The intermediate loss trains within both groups but normalizes them separately, preventing training-only targets from competing with inference candidates. Experiments with a released OneRec model and locally trained Amazon generators show that this competition can harm returned-item ranking. In four prespecified Amazon Video Games comparisons, removing it improved full-target normalized discounted cumulative gain (FT-NDCG) by 7.8--22.2\%; 95\% intervals over users and three of four intervals over training runs excluded zero. A conservative development-set rule selected appended-target training for two of three generators in one held-out category and rejected it for all three in another, avoiding a 1.7\% loss. Candidate completion should therefore be evaluated for each generator rather than applied automatically.
| Comments: | 12 pages, 4 figures, 8 tables |
| Subjects: | Information Retrieval (cs.IR); Machine Learning (cs.LG) |
| ACM classes: | H.3.3 |
| Cite as: | arXiv:2610.10124 [cs.IR] |
| (or arXiv:2610.10124v1 [cs.IR] for this version) | |
| https://doi.org/10.48550/arXiv.2610.10124 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Xuesi Wang [view email]
[v1]
Wed, 7 Oct 2026 14:07:06 UTC (307 KB)
来源：arXiv：cs.LG · arxiv.org