When Is a Relation Worth Trusting? Manipulation Costs, Exposure, and the Trust Trap in Multi-Relational Graph Learning
Shows when an attacked relation is worse to use than to ignore in multi-relational graph learning.
Explainer · on the wire
The explainer for this story is still on the wire.
The claim, the mechanism, and the figures land here once generated.
Source · arXiv wire
arxiv.org/abs/2610.37239cs.LG · cs.AI · cs.CR