GRL-Safety benchmark shows that safety in graph representation learning depends on interactions between method design and specific graph stresses rather than broad method families.
One for all: Towards training one graph model for all classification tasks
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
On the Safety of Graph Representation Learning
GRL-Safety benchmark shows that safety in graph representation learning depends on interactions between method design and specific graph stresses rather than broad method families.