Search

Neural posterior estimation (NPE), a simulation-based computational approach for Bayesian inference, has shown great success in situations where posteriors are intractable or likelihood functions are treated as "black boxes." Existing NPE methods typically rely on normalizing flows, which transform a base distributions into a complex posterior by composing many simple, invertible transformations. But flow-based models, while state of the art for NPE, are known to suffer from several limitations, including training instability and sharp trade-offs between representational power and computational cost. In this work, we demonstrate the effectiveness of conditional diffusions as an alternative to normalizing flows for NPE. Conditional diffusions address many of the challenges faced by flow-based methods. Our results show that, across a highly varied suite of benchmarking problems for NPE architectures, diffusions offer improved stability, superior accuracy, and faster training times, even with simpler, shallower models. These gains persist across a variety of different encoder or "summary network" architectures, as well as in situations where no summary network is required. The code will be publicly available at \url{https://github.com/TianyuCodings/cDiff}.

This paper proposes a Bayesian hierarchical model to characterize the relationship between birth weight and maternal pre-eclampsia across gestation at a large maternity hospital in urban Uganda. Key scientific questions we investigate include: 1) how pre-eclampsia compares to other maternal-fetal covariates as a predictor of birth weight; and 2) whether the impact of pre-eclampsia on birthweight varies across gestation. Our model addresses several key statistical challenges: it correctly encodes the prior medical knowledge that birth weight should vary smoothly and monotonically with gestational age, yet it also avoids assumptions about functional form along with assumptions about how birth weight varies with other covariates. Our model also accounts for the fact that a high proportion (83%) of birth weights in our data set are rounded to the nearest 100 grams. Such extreme data coarsening is rare in maternity hospitals in high resource obstetrics settings but common for data sets collected in low and middle-income countries (LMICs); this introduces a substantial extra layer of uncertainty into the problem and is a major reason why we adopt a Bayesian approach. Our proposed non-parametric regression model, which we call Projective Smooth BART (psBART), builds upon the highly successful Bayesian Additive Regression Tree (BART) framework. This model captures complex nonlinear relationships and interactions, induces smoothness and monotonicity in a single target covariate, and provides a full posterior for uncertainty quantification. The results of our analysis show that pre-eclampsia is a dominant predictor of birth weight in this urban Ugandan setting, and therefore an important risk factor for perinatal mortality.

Ride-sourcing or transportation network companies (TNCs) provide on-demand transportation service for compensation, connecting drivers of personal vehicles with passengers through smartphone applications. In this study, we consider the problem of estimating a spatiotemporally varying probability distribution for the productivity of a TNC driver, using data on more than 1.2 million TNC trips in Austin, Texas. We propose a graph-based smoothing approach that allows for distinct spatial and temporal dynamics, including different degrees of smoothness, spatio-temporal interactions, and interpolation in regions with little or no data. For such a goal, we introduce the Graph-fused Elastic Net (GFEN) and use it in combination with a dyadic tree decomposition for density estimation. In addition, we present an optimization-driven approach for fast point estimates scalable to massive graphs. Bayesian inference and uncertainty quantification with MCMC are also illustrated. The main results demonstrate that the optimization strategy is an effective exploration tool for selecting adequate regularization schemes using Bayesian optimization of the cross-validation loss. Two key empirical findings made possible by our method include: 1) the probability that a TNC driver can expect to earn a living wage in Austin exhibits high variability in space and time, from as low as 25% to as high as 85%; and 2) some drivers suffer considerable "tail risk", with the bottom 10% of the earnings distribution falling below $10 per hour -- grossly below a living wage in Austin for a single adult -- for specific times and locations. All code and data for the paper are publicly available, as a Shiny app for visualizing the results and a software package in Julia for implementing the GFEN.

We introduce Targeted Smooth Bayesian Causal Forests (tsBCF), a nonparametric Bayesian approach for estimating heterogeneous treatment effects which vary smoothly over a single covariate in the observational data setting. The tsBCF method induces smoothness by parameterizing terminal tree nodes with smooth functions, and allows for separate regularization of treatment effects versus prognostic effect of control covariates. Smoothing parameters for prognostic and treatment effects can be chosen to reflect prior knowledge or tuned in a data-dependent way. We use tsBCF to analyze a new clinical protocol for early medical abortion. Our aim is to assess relative effectiveness of simultaneous versus interval administration of mifepristone and misoprostol over the first nine weeks of gestation. The model reflects our expectation that the relative effectiveness varies smoothly over gestation, but not necessarily over other covariates. We demonstrate the performance of the tsBCF method on benchmarking experiments. Software for tsBCF is available at https://github.com/jestarling/tsbcf/.

Many recently developed Bayesian methods have focused on sparse signal detection. However, much less work has been done addressing the natural follow-up question: how to make valid inferences for the magnitude of those signals after selection. Ordinary Bayesian credible intervals suffer from selection bias, owing to the fact that the target of inference is chosen adaptively. Existing Bayesian approaches for correcting this bias produce credible intervals with poor frequentist properties, while existing frequentist approaches require sacrificing the benefits of shrinkage typical in Bayesian methods, resulting in confidence intervals that are needlessly wide. We address this gap by proposing a nonparametric empirical-Bayes approach for constructing optimal selection-adjusted confidence sets. Our method produces confidence sets that are as short as possible on average, while both adjusting for selection and maintaining exact frequentist coverage uniformly over the parameter space. Our main theoretical result establishes an important consistency property of our procedure: that under mild conditions, it asymptotically converges to the results of an oracle-Bayes analysis in which the prior distribution of signal sizes is known exactly. Across a series of examples, the method outperforms existing frequentist techniques for post-selection inference, producing confidence sets that are notably shorter but with the same coverage guarantee.

This paper considers the problem of measuring spatial and temporal variation in driver productivity on ride-sourcing trips. This variation is especially important from a driver's perspective: if a platform's drivers experience systematic disparities in earnings because of variation in their riders' destinations, they may perceive the pricing model as inequitable. This perception can exacerbate search frictions if it leads drivers to avoid locations where they believe they may be assigned "unlucky" fares. To characterize any such systematic disparities in productivity, we develop an analytic framework with three key components. First, we propose a productivity metric that looks two consecutive trips ahead, thus capturing the effect on expected earnings of market conditions at drivers' drop-off locations. Second, we develop a natural experiment by analyzing trips with a common origin but varying destinations, thus isolating purely spatial effects on productivity. Third, we apply a spatial denoising method that allows us to work with raw spatial information exhibiting high levels of noise and sparsity, without having to aggregate data into large, low-resolution spatial zones. By applying our framework to data on more than 1.4 million rides in Austin, Texas, we find significant spatial variation in ride-sourcing driver productivity and search frictions. Drivers at the same location experienced disparities in productivity after being dispatched on trips with different destinations, with origin-based surge pricing increasing these earnings disparities. Our results show that trip distance is the dominant factor in driver productivity: short trips yielded lower productivity, even when ending in areas with high demand. These findings suggest that new pricing strategies are required to minimize random disparities in driver earnings., Comment: 34 pages

This article introduces BART with Targeted Smoothing, or tsBART, a new Bayesian tree-based model for nonparametric regression. The goal of tsBART is to introduce smoothness over a single target covariate t, while not necessarily requiring smoothness over other covariates x. TsBART is based on the Bayesian Additive Regression Trees (BART) model, an ensemble of regression trees. TsBART extends BART by parameterizing each tree's terminal nodes with smooth functions of t, rather than independent scalars. Like BART, tsBART captures complex nonlinear relationships and interactions among the predictors. But unlike BART, tsBART guarantees that the response surface will be smooth in the target covariate. This improves interpretability and helps regularize the estimate. After introducing and benchmarking the tsBART model, we apply it to our motivating example: pregnancy outcomes data from the National Center for Health Statistics. Our aim is to provide patient-specific estimates of stillbirth risk across gestational age (t), based on maternal and fetal risk factors (x). Obstetricians expect stillbirth risk to vary smoothly over gestational age, but not necessarily over other covariates, and tsBART has been designed precisely to reflect this structural knowledge. The results of our analysis show the clear superiority of the tsBART model for quantifying stillbirth risk, thereby providing patients and doctors with better information for managing the risk of perinatal mortality. All methods described here are implemented in the R package tsbart.

Individuals in low socioeconomic brackets are considered at-risk for developing influenza-related complications and often exhibit higher than average influenza-related hospitalization rates. This disparity has been attributed to various factors, including restricted access to preventative and therapeutic health care, limited sick leave, and household structure. Adequate influenza surveillance in these at-risk populations is a critical precursor to accurate risk assessments and effective intervention. However, the United States of America's primary national influenza surveillance system (ILINet) monitors outpatient healthcare providers, which may be largely inaccessible to lower socioeconomic populations. Recent initiatives to incorporate internet-source and hospital electronic medical records data into surveillance systems seek to improve the timeliness, coverage, and accuracy of outbreak detection and situational awareness. Here, we use a flexible statistical framework for integrating multiple surveillance data sources to evaluate the adequacy of traditional (ILINet) and next generation (BioSense 2.0 and Google Flu Trends) data for situational awareness of influenza across poverty levels. We find that zip codes in the highest poverty quartile are a critical blind-spot for ILINet that the integration of next generation data fails to ameliorate.

We consider the problem of estimating a regression function in the common situation where the number of features is small, where interpretability of the model is a high priority, and where simple linear or additive models fail to provide adequate performance. To address this problem, we present Maximum Variance Total Variation denoising (MVTV), an approach that is conceptually related both to CART and to the more recent CRISP algorithm, a state-of-the-art alternative method for interpretable nonlinear regression. MVTV divides the feature space into blocks of constant value and fits the value of all blocks jointly via a convex optimization routine. Our method is fully data-adaptive, in that it incorporates highly robust routines for tuning all hyperparameters automatically. We compare our approach against CART and CRISP via both a complexity-accuracy tradeoff metric and a human study, demonstrating that that MVTV is a more powerful and interpretable method., Comment: 4 pages, 1 figure presented at 2017 ICML Workshop on Human Interpretability in Machine Learning (WHI 2017), Sydney, NSW, Australia

We consider the problem of estimating a regression function in the common situation where the number of features is small, where interpretability of the model is a high priority, and where simple linear or additive models fail to provide adequate performance. To address this problem, we present GapTV, an approach that is conceptually related both to CART and to the more recent CRISP algorithm, a state-of-the-art alternative method for interpretable nonlinear regression. GapTV divides the feature space into blocks of constant value and fits the value of all blocks jointly via a convex optimization routine. Our method is fully data-adaptive, in that it incorporates highly robust routines for tuning all hyperparameters automatically. We compare our approach against CART and CRISP and demonstrate that GapTV finds a much better trade-off between accuracy and interpretability.

We present an approach to deep estimation of discrete conditional probability distributions. Such models have several applications, including generative modeling of audio, image, and video data. Our approach combines two main techniques: dyadic partitioning and graph-based smoothing of the discrete space. By recursively decomposing each dimension into a series of binary splits and smoothing over the resulting distribution using graph-based trend filtering, we impose a strict structure to the model and achieve much higher sample efficiency. We demonstrate the advantages of our model through a series of benchmarks on both synthetic and real-world datasets, in some cases reducing the error by nearly half in comparison to other popular methods in the literature. All of our models are implemented in Tensorflow and publicly available at https://github.com/tansey/sdp .