内容提示: MULTI-VIEW CONTRASTIVE LEARNING FOR ONLINE KNOWLEDGE DISTILLATIONChuanguang Yang ?† Zhulin An?‡∗Yongjun Xu ?‡?Institute of Computing Technology (ICT), Chinese Academy of Sciences (CAS), Beijing, China† University of Chinese Academy of Sciences, Beijing, China‡Xiamen Data ...
Multi-view Contrastive Learning 多视图CL旨在突出单个序列和全局图视图之间的关系。很自然,同一个用户的序列视图和图形视图用户表示应该比其他用户更接近,因为它们反映了同一个用户的偏好。 Multi-view CL Behavior Distinction Contrastive Learning 以上两个CL(Multi-behavior/Multi-view)任务突出了用户的多个行为和视图...
AdaMCL: Adaptive Fusion Multi-View Contrastive Learning for Collaborative Filtering Abstract 大多数基于CL的方法只利用原始的用户-项目交互图来构造CL任务,缺乏对高阶信息的显示利用。而且即使是使用高阶信息的基于CL的方法,高阶信息的接收字段也是固定的,没有考虑到节点之间的差异,于是我们提出了一种新的自...
Contrastive Learning 对比学习在视觉表示学习、自然语言处理和图神经网络中取得了令人印象深刻的成就。最近,一些研究将对比学习引入推荐系统,例如SGL通过节点自我辨别为基于GCN的推荐模型提供辅助信号。SEPT设计了一个社会感知的自监督框架,从用户-项目图和社会图中学习区分信号。一些工作也将对比学习引入到序列推荐中,S^{...
Finally, three multi-view contrastive learning objectives are combined for better IDRR. Our proposed method only slightly increases training time and introduces small additional parameters. Experimental results on PDTB 2. 0 show that our method achieves the state-of...
论文阅读组会版#Heterogeneous Graph Contrastive Multi-view Learning(HGCML)82 0 2024-01-25 16:55:48 未经作者授权,禁止转载 您当前的浏览器不支持 HTML5 播放器 请更换浏览器再试试哦~2 3 3 分享 记录一下阅读的论文 知识 校园学习 研究生组会 论文 GNN 论文阅读 ...
MCLET is composed of three modules: i) Multi-view Generation and Encoder module, which encodes structured information from entity-type, entity-cluster and cluster-type views; ii) Cross-view Contrastive Learning module, which encourages different views to collaboratively improve view-specific ...
“Multi-view heterogeneous graph contrastive learning” section describes the implementation of the multi-view contrastive learning for heterogeneous network embedding; “Multi-view heterogeneous graph contrastive learning” section provides the experimental results; finally, the research is summarized in “...
Multi-level Contrastive Learning Framework for Sequential Recommendation 用于序列推荐的多层次对比学习框架 来源:CIKM 2022 摘要:顺序推荐(SR)旨在通过了解用户连续的历史行为来预测用户的后续行为。最近,一些SR的方法致力于缓解数据稀疏性问题(即训练的有限监督信号),考虑使用对比学习将自监督信号纳入SR。尽管这样的做法...
What Should Not Be Contrastive in Contrastive Learning学习笔记 What Should Not Be Contrastive in Contrastive Learning 引入了一个对比学习框架,该框架不需要事先特定的,与任务相关的不变性的知识,模型学会捕捉通过构建单独的视觉表示的可变和不变因素嵌入到空间,除了扩充之外,每个空间都是不变的。 数据增强的引入...