Adversarially Training MCMC with Non-Volume-Preserving Flows
Sign Up to like & getrecommendations! Published in 2022 at "Entropy"
DOI: 10.3390/e24030415
Abstract: Recently, flow models parameterized by neural networks have been used to design efficient Markov chain Monte Carlo (MCMC) transition kernels. However, inefficient utilization of gradient information of the target distribution or the use of volume-preserving… read more here.
Keywords: non volume; preserving flows; transition kernels; transition ... See more keywords