Neural Methods for Amortised Parameter Inference

Simulation-based methods for making statistical inference have evolved dramatically over the past 50 years, keeping pace with technological advancements. The field is undergoing a new revolution as it embraces the representational capacity of neural networks, optimisation libraries, and graphics processing units for learning complex mappings between data and inferential targets. The resulting tools are amortised, in the sense that they allow inference to be made quickly through fast feedforward operations. In this article we review recent progress made in the context of point estimation, approximate Bayesian inference, the automatic construction of summary statistics, and likelihood approximation. The review also covers available software, and includes a simple illustration to showcase the wide array of tools available for amortised inference and the benefits they offer over state-of-the-art Markov chain Monte Carlo methods. The article concludes with an overview of relevant topics and an outlook on future research directions.

Spatial Bayesian Neural Networks

Statistical models for spatial processes play a central role in statistical analyses of spatial data. Yet, it is the simple, interpretable, and well understood models that are routinely employed even though, as is revealed through prior and posterior predictive checks, these can poorly characterise the spatial heterogeneity in the underlying process of interest. Here, we propose a new, flexible class of spatial-process models, which we refer to as spatial Bayesian neural networks (SBNNs). An SBNN leverages the representational capacity of a Bayesian neural network; it is tailored to a spatial setting by incorporating a spatial "embedding layer" into the network and, possibly, spatially-varying network parameters. An SBNN is calibrated by matching its finite-dimensional distribution at locations on a fine gridding of space to that of a target process of interest. That process could be easy to simulate from or we have many realisations from it. We propose several variants of SBNNs, most of which are able to match the finite-dimensional distribution of the target process at the selected grid better than conventional BNNs of similar complexity. We also show that a single SBNN can be used to represent a variety of spatial processes often used in practice, such as Gaussian processes and lognormal processes. We briefly discuss the tools that could be used to make inference with SBNNs, and we conclude with a discussion of their advantages and limitations.

Neural Bayes Estimators for Irregular Spatial Data using Graph Neural Networks

Neural Bayes estimators are neural networks that approximate Bayes estimators in a fast and likelihood-free manner. They are appealing to use with spatial models and data, where estimation is often a computational bottleneck. However, neural Bayes estimators in spatial applications have, to date, been restricted to data collected over a regular grid. These estimators are also currently dependent on a prescribed set of spatial locations, which means that the neural network needs to be re-trained for new data sets; this renders them impractical in many applications and impedes their widespread adoption. In this work, we employ graph neural networks to tackle the important problem of parameter estimation from data collected over arbitrary spatial locations. In addition to extending neural Bayes estimation to irregular spatial data, our architecture leads to substantial computational benefits, since the estimator can be used with any arrangement or number of locations and independent replicates, thus amortising the cost of training for a given spatial model. We also facilitate fast uncertainty quantification by training an accompanying neural Bayes estimator that approximates a set of marginal posterior quantiles. We illustrate our methodology on Gaussian and max-stable processes. Finally, we showcase our methodology in a global sea-surface temperature application, where we estimate the parameters of a Gaussian process model in 2,161 regions, each containing thousands of irregularly-spaced data points, in just a few minutes with a single graphics processing unit.

Likelihood-free neural Bayes estimators for censored inference with peaks-over-threshold models

Inference for spatial extremal dependence models can be computationally burdensome in moderate-to-high dimensions due to their reliance on intractable and/or censored likelihoods. Exploiting recent advances in likelihood-free inference with neural Bayes estimators (that is, neural estimators that target Bayes estimators), we develop a novel approach to construct highly efficient estimators for censored peaks-over-threshold models by encoding censoring information in the neural network architecture. Our new method provides a paradigm shift that challenges traditional censored likelihood-based inference for spatial extremes. Our simulation studies highlight significant gains in both computational and statistical efficiency, relative to competing likelihood-based approaches, when applying our novel estimators for inference of popular extremal dependence models, such as max-stable, $r$-Pareto, and random scale mixture processes. We also illustrate that it is possible to train a single estimator for a general censoring level, obviating the need to retrain when the censoring level is changed. We illustrate the efficacy of our estimators by making fast inference on hundreds-of-thousands of high-dimensional spatial extremal dependence models to assess particulate matter 2.5 microns or less in diameter (PM2.5) concentration over the whole of Saudi Arabia.

Fast Optimal Estimation with Intractable Models using Permutation-Invariant Neural Networks

Neural networks have recently shown promise for likelihood-free inference, providing orders-of-magnitude speed-ups over classical methods. However, current implementations are suboptimal when estimating parameters from independent replicates. In this paper, we use a decision-theoretic framework to argue that permutation-invariant neural networks are ideally placed for constructing Bayes estimators for arbitrary models, provided that simulation from these models is straightforward. We illustrate the potential of these estimators on both conventional spatial models, as well as highly parameterised spatial-extremes models, and show that they considerably outperform neural estimators that do not account for replication appropriately in their network design. At the same time, they are highly competitive and much faster than traditional likelihood-based estimators. We apply our estimator on a spatial analysis of sea-surface temperature in the Red Sea where, after training, we obtain parameter estimates, and uncertainty quantification of the estimates via bootstrap sampling, from hundreds of spatial fields in a fraction of a second.

Statistical Deep Learning for Spatial and Spatio-Temporal Data

Deep neural network models have become ubiquitous in recent years, and have been applied to nearly all areas of science, engineering, and industry. These models are particularly useful for data that have strong dependencies in space (e.g., images) and time (e.g., sequences). Indeed, deep models have also been extensively used by the statistical community to model spatial and spatio-temporal data through, for example, the use of multi-level Bayesian hierarchical models and deep Gaussian processes. In this review, we first present an overview of traditional statistical and machine learning perspectives for modeling spatial and spatio-temporal data, and then focus on a variety of hybrid models that have recently been developed for latent process, data, and parameter specifications. These hybrid models integrate statistical modeling ideas with deep neural network models in order to take advantage of the strengths of each modeling paradigm. We conclude by giving an overview of computational technologies that have proven useful for these hybrid models, and with a brief discussion on future research directions.

Spherical Poisson Point Process Intensity Function Modeling and Estimation with Measure Transport

Recent years have seen an increased interest in the application of methods and techniques commonly associated with machine learning and artificial intelligence to spatial statistics. Here, in a celebration of the ten-year anniversary of the journal Spatial Statistics, we bring together normalizing flows, commonly used for density function estimation in machine learning, and spherical point processes, a topic of particular interest to the journal's readership, to present a new approach for modeling non-homogeneous Poisson process intensity functions on the sphere. The central idea of this framework is to build, and estimate, a flexible bijective map that transforms the underlying intensity function of interest on the sphere into a simpler, reference, intensity function, also on the sphere. Map estimation can be done efficiently using automatic differentiation and stochastic gradient descent, and uncertainty quantification can be done straightforwardly via nonparametric bootstrap. We investigate the viability of the proposed method in a simulation study, and illustrate its use in a proof-of-concept study where we model the intensity of cyclone events in the North Pacific Ocean. Our experiments reveal that normalizing flows present a flexible and straightforward way to model intensity functions on spheres, but that their potential to yield a good fit depends on the architecture of the bijective map, which can be difficult to establish in practice.

Emulation of greenhouse-gas sensitivities using variational autoencoders

Flux inversion is the process by which sources and sinks of a gas are identified from observations of gas mole fraction. The inversion often involves running a Lagrangian particle dispersion model (LPDM) to generate sensitivities between observations and fluxes over a spatial domain of interest. The LPDM must be run backward in time for every gas measurement, and this can be computationally prohibitive. To address this problem, here we develop a novel spatio-temporal emulator for LPDM sensitivities that is built using a convolutional variational autoencoder (CVAE). With the encoder segment of the CVAE, we obtain approximate (variational) posterior distributions over latent variables in a low-dimensional space. We then use a spatio-temporal Gaussian process emulator on the low-dimensional space to emulate new variables at prediction locations and time points. Emulated variables are then passed through the decoder segment of the CVAE to yield emulated sensitivities. We show that our CVAE-based emulator outperforms the more traditional emulator built using empirical orthogonal functions and that it can be used with different LPDMs. We conclude that our emulation-based approach can be used to reliably reduce the computing time needed to generate LPDM outputs for use in high-resolution flux inversions.

Deep Integro-Difference Equation Models for Spatio-Temporal Forecasting

Integro-difference equation (IDE) models describe the conditional dependence between the spatial process at a future time point and the process at the present time point through an integral operator. Nonlinearity or temporal dependence in the dynamics is often captured by allowing the operator parameters to vary temporally, or by re-fitting a model with a temporally-invariant linear operator at each time point in a sliding window. Both procedures tend to be excellent for prediction purposes over small time horizons, but are generally time-consuming and, crucially, do not provide a global prior model for the temporally-varying dynamics that is realistic. Here, we tackle these two issues by using a deep convolution neural network (CNN) in a hierarchical statistical IDE framework, where the CNN is designed to extract process dynamics from the process' most recent behaviour. Once the CNN is fitted, probabilistic forecasting can be done extremely quickly online using an ensemble Kalman filter with no requirement for repeated parameter estimation. We conduct an experiment where we train the model using 13 years of daily sea-surface temperature data in the North Atlantic Ocean. Forecasts are seen to be accurate and calibrated. A key advantage of our approach is that the CNN provides a global prior model for the dynamics that is realistic, interpretable, and computationally efficient. We show the versatility of the approach by successfully producing 10-minute nowcasts of weather radar reflectivities in Sydney using the same model that was trained on daily sea-surface temperature data in the North Atlantic Ocean.

Deep Compositional Spatial Models

Nonstationary, anisotropic spatial processes are often used when modelling, analysing and predicting complex environmental phenomena. One such class of processes considers a stationary, isotropic process on a warped spatial domain. The warping function is generally difficult to fit and not constrained to be bijective, often resulting in 'space-folding.' Here, we propose modelling a bijective warping function through a composition of multiple elemental bijective functions in a deep-learning framework. We consider two cases; first, when these functions are known up to some weights that need to be estimated, and, second, when the weights in each layer are random. Inspired by recent methodological and technological advances in deep learning and deep Gaussian processes, we employ approximate Bayesian methods to make inference with these models using graphical processing units. Through simulation studies in one and two dimensions we show that the deep compositional spatial models are quick to fit, and are able to provide better predictions and uncertainty quantification than other deep stochastic models of similar complexity. We also show their remarkable capacity to model highly nonstationary, anisotropic spatial data using radiances from the MODIS instrument aboard the Aqua satellite.