深度稳定多兴趣学习用于序列推荐中的分布外情况（英文版）

英文标题：Deep Stable Multi-Interest Learning for Out-of-distribution Sequential Recommendation中文摘要：该研究提出了一种名为 DESMIL 的新型多兴趣网络，通过使用注意模块提取多个兴趣，并基于 Hilbert-Schmidt 独立性标准（HSIC）估计加权相关性损失，从而最小化提取的兴趣之间的相关性，解决了现有多兴趣推荐模型没有考虑兴趣分布更改可能导致的泛化问题。英文摘要：Recently, multi-interest models, which extract interests of a user asmultiple representation vectors, have shown promising performances forsequential recommendation. However, none of existing multi-interestrecommendation models consider the Out-Of-Distribution (OOD)