Demystifying Relational Latent Representations
This work addresses the black-box nature of latent representations in relational learning, offering incremental improvements in interpretability and efficiency for researchers and practitioners in machine learning.
The paper tackles the problem of interpretability in relational latent representations by demonstrating that clustering-based latent features are interpretable, capture data properties, and identify label-matching regions, while also showing that many features are redundant and can be pruned without significant performance loss.
Latent features learned by deep learning approaches have proven to be a powerful tool for machine learning. They serve as a data abstraction that makes learning easier by capturing regularities in data explicitly. Their benefits motivated their adaptation to relational learning context. In our previous work, we introduce an approach that learns relational latent features by means of clustering instances and their relations. The major drawback of latent representations is that they are often black-box and difficult to interpret. This work addresses these issues and shows that (1) latent features created by clustering are interpretable and capture interesting properties of data; (2) they identify local regions of instances that match well with the label, which partially explains their benefit; and (3) although the number of latent features generated by this approach is large, often many of them are highly redundant and can be removed without hurting performance much.