我在做实验的时候,所涉及的场景确实属于target本身已经能够训得不错,需要一些cross-domain锦上添花。如果说target本身比source少太多,有几个办法: ①对source样本采样,然后source_batch_size设置小于target_batch_size,或许可减少梯度带偏的影响。②将两边的embed_table合并,既然target本身已经不足以收敛了,那么合并...
跨域推荐中的关键问题是如何根据用户的将他们的偏好有效地从一个域迁移到另一个相关域。受基于评论的推荐的进步的启发,我们提议对用户偏好的迁移进行aspect级的建模,aspect从评论中得出。为此,我们提出了一种通过aspect迁移网络针对冷启动用户的跨域推荐框架(命名为CATN)。 CATN旨在从他们的评论文档中为每个用户和每个...
跨域推荐。跨域推荐(Cross-domain recommendation, CDR)是利用其他域的辅助用户行为来缓解数据稀疏性问题的代表性方法之一[24,53]。经典的CDR方法一般采用多任务学习[52]、对齐约束[20,35]和对比学习[41]对跨领域知识迁移进行建模。跨域顺序推荐(CDSR)更多关注用户在CDR中的多域时间顺序行为[2,3,10,17,39,46]。...
[ICDE 2022]Cross-Domain Recommendation to Cold-Start Users via Variational Information Bottleneck information-bottleneckcross-domain-recommendation UpdatedDec 7, 2022 Python Chain123/RecGURU Star27 Code Issues Pull requests The source code and dataset for the RecGURU paper (WSDM 2022) ...
Cross-domain recommendationCollaborative filteringCold-start usersUser similarityLatent feature mappingCollaborative filtering (CF) is a widely adopted technique in recommender systems. Traditional CF models mainly focus on predicting the user preference to items in a single domain, such as the movie ...
随着人们不可避免地跨越多个领域或各种平台与物品进行交互,跨领域推荐( Cross-domain Recommendation,CDR )得到了越来越多的关注。然而,日益增长的隐私问题限制了现有CDR模型的实际应用,因为它们假设不同领域之间的全部或部分数据是可访问的。 最近关于隐私感知CDR模型的研究忽略了来自多个领域数据的异构性,无法在跨领域推...
To address the problem of sparse data and cold-start when facing new users and items in the single-domain recommendation, cross-domain recommendation has gradually become a hot topic in the recommendation system. This method enhances target domain recommendation performance by incorporating relevant info...
Cross-Domain Recommendation (CDR) aims to solve data sparsity and cold-start problems by utilizing a relatively information-rich source domain to improve the recommendation performance of the data-sparse target domain. However, most existing approaches rely on the assumption of centralized storage of ...
内容提示: Federated Graph Learning for Cross-DomainRecommendationZiqi Yang 1,2 , Zhaopeng Peng 1,2 , Zihui Wang 1,2 , Jianzhong Qi 3 , Chaochao Chen 4 ,Weike Pan 5 , Chenglu Wen 1,2 , Cheng Wang 1,2 , Xiaoliang Fan 1,2 ∗1 Fujian Key Laboratory of Sensing and Computing for ...
Transfer-Meta Framework for Cross-domain Recommendation to Cold-Start Users阅读笔记 动机 本文是2021年SIGIR上的一篇论文。本文主要针对的是冷启动问题中的跨域推荐问题,目前常用的方法是EMCDR,但是这个方法很大局限性,它仅在重叠的用户上学习,这样学到的模型会偏向于这些重叠用户,针对以上问题,本文提出了TMCDR方法...