Nonlinear MCMC for Bayesian Machine Learning
This work addresses sampling problems in Bayesian machine learning, but it appears incremental as it builds on an existing nonlinear MCMC technique.
The paper tackles the challenge of applying nonlinear MCMC techniques to Bayesian machine learning by providing convergence guarantees and demonstrating its use on a Bayesian neural network with CIFAR10, though no concrete numerical results are reported.
We explore the application of a nonlinear MCMC technique first introduced in [1] to problems in Bayesian machine learning. We provide a convergence guarantee in total variation that uses novel results for long-time convergence and large-particle ("propagation of chaos") convergence. We apply this nonlinear MCMC technique to sampling problems including a Bayesian neural network on CIFAR10.