kTULA achieves a non-asymptotic KL convergence of order lambda^(2-epsilon) for non-log-concave targets with super-linear log-gradients under a Log-Sobolev inequality, improving prior order-lambda KL bounds.
On Stochastic Gradient Langevin Dynamics with Dependent Data Streams: The Fully Nonconvex Case.SIAM Journal on Mathematics of Data Science, 3(3):959–986, 2021
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
math.ST 1years
2025 1verdicts
ACCEPT 1representative citing papers
citing papers explorer
-
kTULA: A Langevin sampling algorithm with improved KL bounds under super-linear log-gradients
kTULA achieves a non-asymptotic KL convergence of order lambda^(2-epsilon) for non-log-concave targets with super-linear log-gradients under a Log-Sobolev inequality, improving prior order-lambda KL bounds.