Scalable Feature Learning on Huge Knowledge Graphs for Downstream Machine Learning

ArXi:2507.00965v3 Announce Type: replace Many machine learning tasks can benefit from external knowledge. Large knowledge graphs such knowledge, and embedding methods can be used to distill it into ready-to-use vector representations for downstream applications. For this purpose, current models have. however. two limitations: they are primarily optimized for link prediction, via local contrastive learning, and their application to the largest graphs requires significant engineering effort due to GPU memory limits. To address these, we.