1 paper · 1 filter
Ziyue Wang, Takafumi Kanamori
We study score learning for reflected diffusion on bounded domains. Reflection keeps trajectories feasible but does not ensure that the learned score satisfies the boundary behavio…