A stochastic-interpolant sampler that solves HJB PDEs via FBSDEs with neural networks, giving finite-time Gaussian-to-target transport with mixed empirical success.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Sampling in High-Dimensions using Stochastic Interpolants and Forward-Backward Stochastic Differential Equations
A stochastic-interpolant sampler that solves HJB PDEs via FBSDEs with neural networks, giving finite-time Gaussian-to-target transport with mixed empirical success.