A JKO-based method recovers free-energy functionals for discrete graph diffusion under the WK metric, enabling fast quadratic-loss training without sample trajectories.
transport cost
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
-
Learning Discrete Diffusion of Graphs via Free-Energy Gradient Flows
A JKO-based method recovers free-energy functionals for discrete graph diffusion under the WK metric, enabling fast quadratic-loss training without sample trajectories.