Relational Constraints for Metric Learning on Relational Data
This work addresses metric learning for relational data, which is a domain-specific problem, and is incremental as it builds on existing methods like ITML and LSML.
The paper tackles the problem of metric learning for relational data by proposing an algorithm that incorporates both topological structure and supervised labels, using a link-strength function based on common parents. Experimental results on real-world datasets show that integrating relational information improves metric quality, with specific gains observed when combined with ITML and LSML methods.
Most of metric learning approaches are dedicated to be applied on data described by feature vectors, with some notable exceptions such as times series, trees or graphs. The objective of this paper is to propose a metric learning algorithm that specifically considers relational data. The proposed approach can take benefit from both the topological structure of the data and supervised labels. For selecting relative constraints representing the relational information, we introduce a link-strength function that measures the strength of relationship links between entities by the side-information of their common parents. We show the performance of the proposed method with two different classical metric learning algorithms, which are ITML (Information Theoretic Metric Learning) and LSML (Least Squares Metric Learning), and test on several real-world datasets. Experimental results show that using relational information improves the quality of the learned metric.