An introduction to sampling via measure transport

By Youssef M. Marzouk, Tarek Moselhy, Matthew D. Parno, Alessio Spantini. 2016

97 citations recorded by the source corpus.

We present the fundamentals of a measure transport approach to sampling. The idea is to construct a deterministic coupling---i.e., a transport map---between a complex "target" probability measure of interest and a simpler reference measure. Given a transport map, one can generate arbitrarily many independent and unweighted samples from the target simply by pushing forward reference samples through the map. We consider two different and complementary scenarios: first, when only evaluations of the unnormalized target density are available, and second, when the target distribution is known only through a finite collection of samples. We show that in both settings the desired transports can be characterized as the solutions of variational problems. We then address practical issues associated with the optimization--based construction of transports: choosing finite-dimensional parameterizations of the map, enforcing monotonicity, quantifying the error of approximate transports, and refining approximate transports by enriching the corresponding approximation spaces. Approximate transports can also be used to "Gaussianize" complex distributions and thus precondition conventional asymptotically exact sampling schemes. We place the measure transport approach in broader context, describing connections with other optimization--based samplers, with inference and density estimation schemes using optimal transport, and with alternative transformation--based approaches to simulation. We also sketch current work aimed at the construction of transport maps in high dimensions, exploiting essential features of the target distribution (e.g., conditional independence, low-rank structure). The approaches and algorithms presented here have direct applications to Bayesian computation and to broader problems of stochastic simulation.

View Paper: https://www.semanticscholar.org/paper/8ed71565641df500687006d3c3b286685ce0d742

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

Stein Variational Gradient Descent: A General Purpose Bayesian Inference Algorithm (2016, background): “[14], which also considers transforms parameterized in special ways to ensure the invertible and the computational tractability of the Jacobian matrix.”
Adversarial Generator-Encoder Networks (2017, background): “For a detailed discussion on this topic please refer to (Villani, 2008; Marzouk et al., 2016).”
Bayesian Learning with Wasserstein Barycenters (2018, methodology): “In the parametric setting, we can use e ﬃ cient Markov Chain Monte Carlo (MCMC) techniques [22] or transport sampling procedures [17, 34, 24, 30] to generate models m i sampled from Π n or Π .”
It Takes (Only) Two: Adversarial Generator-Encoder Networks (2017, background): “For a detailed discussion on this topic please refer to [20, 27].”