Computation of heterogeneous object co-embeddings from relational measurements

详细信息    查看全文

• 作者：Yu Wu^a ; ^{yu.wu@liverpool.ac.uk}Author Vitae ; Tingting Mu^b ; ^{tingtingmu@me.com}Author Vitae ; Panos Liatsis^c ; ^{pliatsis@pi.ac.ae}Author Vitae ; John Y. Goulermas^a ; ^{j.y.goulermas@liverpool.ac.uk}Author Vitae
• 关键词：Co-embedding generation ; Relational information ; Heterogeneous object analysis ; Joint space projection.
• 刊名：Pattern Recognition
• 出版年：2017
• 出版时间：May 2017
• 年：2017
• 卷：65
• 期：Complete
• 页码：146-163
• 全文大小：3942 K
• 卷排序：65

文摘

Analysis of arbitrary relational information between heterogeneous objects. Explicit generation of co-embeddings directly from relational measurements. High efficiency due to simple matrix decomposition and a small set of parameters. Simple scheme to implement the model identification of the parameters. Multiple comparisons with synthetic and real-world datasets, and existing algorithms.