Inference Networks for Sequential Monte Carlo in Graphical Models

By Brooks Paige, Frank D. Wood. 2016

116 citations recorded by the source corpus.

We introduce a new approach for amortizing inference in directed graphical models by learning heuristic approximations to stochastic inverses, designed specifically for use as proposal distributions in sequential Monte Carlo methods. We describe a procedure for constructing and learning a structured neural network which represents an inverse factorization of the graphical model, resulting in a conditional density estimator that takes as input particular values of the observed random variables, and returns an approximation to the distribution of the latent variables. This recognition model can be learned offline, independent from any particular dataset, prior to performing inference. The output of these networks can be used as automatically-learned high-quality proposal distributions to accelerate sequential Monte Carlo across a diverse range of problem settings.

Subfields: Artificial Intelligence, Probabilistic & Bayesian Methods

How researchers see this paper

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

In Artificial Intelligence

In Probabilistic & Bayesian Methods

Citations in the corpus

An Introduction to Probabilistic Programming (2018, methodology): “In (Paige and Wood, 2016) a masked auto-regression density estimator was used.”
Masked Autoregressive Flow for Density Estimation (2017, methodology): “The second step would be to approximate Equation (13) with Monte Carlo, using samples {xn} drawn from πx(x), as follows: DKL(pu(u) ‖πu(u)) = Eπx(x) ( log πx(x)− log πu ( f−1(x) ) − log ∣∣∣∣det(∂f−1∂x )∣∣∣∣) (18) ≈ 1 N ∑ n ( log πx(xn)− log πu ( f−1(xn) ) − log ∣∣∣∣det(∂f−1∂x )∣∣∣∣).”
Learning to Draw Samples: With Application to Amortized MLE for Generative Adversarial Learning (2016, methodology): “…2014) has been recently applied in various domains of probabilistic reasoning, including both amortized variational inference (e.g., Kingma & Welling, 2013; Rezende & Mohamed, 2015a), and data-driven proposals for (sequential) Monte Carlo methods (e.g., Paige & Wood, 2016), to name only a few.”
Inference Compilation and Universal Probabilistic Programming (2016, methodology): “We obtain K proposal execution traces x k := x k 1: T k (possibly in parallel) to which we assign weights for k = 1 , . . . , K with T k denoting the length of the k th proposal execution trace.”
Translating Neuralese (2017, background): “While in general neither humans nor DCP agents compute explicit representations of this posterior, past work has found that both humans and suitably-trained neural networks can be modeled as Bayesian reasoners (Frank et al., 2009; Paige and Wood, 2016).”
Fast $\epsilon$-free Inference of Simulation Models with Bayesian Conditional Density Estimation (2016, methodology): “Neural density estimators have been employed in learning proposal distributions for importance sampling [24] and Sequential Monte Carlo [11, 23].”
Meta-Learning an Inference Algorithm for Probabilistic Programs (2021, background): “Amortised inference and inference compilation (Gershman & Goodman, 2014; Le et al., 2017; Paige & Wood, 2016; Stuhlmüller et al., 2013; Kingma & Welling, 2013; Mnih & Gregor, 2014; Rezende et al., 2014; Ritchie et al., 2016; Marino et al., 2018) are closely related to our approach in that they also…”
Likelihood-free inference with an improved cross-entropy estimator (2018, unknown): “+ (1− yi) ∣∣∣∣ 1 r(xi, zi|θ0, θ1) − 1 r̂(x|θ0, θ1) ∣∣∣∣2] (13) is minimized.”
Tighter Variational Bounds are Not Necessarily Better (2018, methodology): “Variational bounds are also used in amortized and traditional Bayesian inference contexts (Hoffman et al., 2013; Ranganath et al., 2014; Paige & Wood, 2016; Le et al., 2017), for which the generative model is ﬁxed and the inference network is the primary target for the training.”
Mining for Dark Matter Substructure: Inferring Subhalo Population Properties from Strong Lenses with Machine Learning (2019, background): “…ratio function, or posterior(Fan et al. 2012; Dinh et al. 2014, 2016; Cranmer et al. 2015; Germain et al. 2015; Jimenez Rezende & Mohamed 2015; Paige & Wood 2016; Papamakarios & Murray 2016; Thomas et al. 2016; Uria et al. 2016; van den Oord et al. 2016b, 2016c, 2016a; Gutmann et al. 2017;…”
A Universal Marginalizer for Amortized Inference in Generative Models (2017, methodology): “Amortized inference has been popular for Sequential Monte Carlo and has been used to learn in advance either parameters [Gu et al., 2015] or a discriminative model which provides conditional density estimates [Morris, 2001, Paige and Wood, 2016].”
Learning Approximate Inference Networks for Structured Prediction (2018, methodology): “Since we train a single inference network for an entire dataset, our approach is also related to “amortized inference” (Srikumar et al., 2012; Gershman & Goodman, 2014; Paige & Wood, 2016; Chang et al., 2015).”
Dynamical Sampling with Langevin Normalization Flows (2019, unknown): “Recently, neural density estimators are widely used in the approximation of data distributions [32,33] and variational inference [27].”
Sequential Neural Likelihood: Fast Likelihood-free Inference with Autoregressive Flows (2018, methodology): “Beyond SNPE, conditional neural density estimators have been used to learn the posterior from simulations in graphical models [53, 58], and universal probabilistic programs [38].”
Deep Amortized Inference for Probabilistic Programs (2016, background): “And for models in which there is a natural hierarchy between the latent variables and the observed variables, having the guide run ‘backwards’ from the observed variables to the top-most latents has been shown to be useful [28, 22, 20].”