اشترك بالحزمة الذهبية واحصل على وصول غير محدود شمرا أكاديميا
تسجيل مستخدم جديدFast Algorithms and Theory for High-Dimensional Bayesian Varying Coefficient Models
62
0
0.0
(
0
)
اسأل ChatGPT حول البحث
ﻻ يوجد ملخص باللغة العربية
Nonparametric varying coefficient (NVC) models are useful for modeling time-varying effects on responses that are measured repeatedly. In this paper, we introduce the nonparametric varying coefficient spike-and-slab lasso (NVC-SSL) for Bayesian estimation and variable selection in NVC models. The NVC-SSL simultaneously selects and estimates the significant varying coefficients, while also accounting for temporal correlations. Our model can be implemented using a computationally efficient expectation-maximization (EM) algorithm. We also employ a simple method to make our model robust to misspecification of the temporal correlation structure. In contrast to frequentist approaches, little is known about the large-sample properties for Bayesian NVC models when the dimension of the covariates $p$ grows much faster than sample size $n$. In this paper, we derive posterior contraction rates for the NVC-SSL model when $p gg n$ under both correct specification and misspecification of the temporal correlation structure. Thus, our results are derived under weaker assumptions than those seen in other high-dimensional NVC models which assume independent and identically distributed (iid) random errors. Finally, we illustrate our methodology through simulation studies and data analysis. Our method is implemented in the publicly available R package NVCSSL.
قيم البحث
اقرأ أيضاً
This paper investigates the integration of gradient boosted decision trees and varying coefficient models. We introduce the tree boosted varying coefficient framework which justifies the implementation of decision tree boosting as the nonparametric e
ffect modifiers in varying coefficient models. This framework requires no structural assumptions in the space containing the varying coefficient covariates, is easy to implement, and keeps a balance between model complexity and interpretability. To provide statistical guarantees, we prove the asymptotic consistency of the proposed method under the regression settings with $L^2$ loss. We further conduct a thorough empirical study to show that the proposed method is capable of providing accurate predictions as well as intelligible visual explanations.
There has been growing interest in the AI community for precise uncertainty quantification. Conditional density models f(y|x), where x represents potentially high-dimensional features, are an integral part of uncertainty quantification in prediction
and Bayesian inference. However, it is challenging to assess conditional density estimates and gain insight into modes of failure. While existing diagnostic tools can determine whether an approximated conditional density is compatible overall with a data sample, they lack a principled framework for identifying, locating, and interpreting the nature of statistically significant discrepancies over the entire feature space. In this paper, we present rigorous and easy-to-interpret diagnostics such as (i) the Local Coverage Test (LCT), which distinguishes an arbitrarily misspecified model from the true conditional density of the sample, and (ii) Amortized Local P-P plots (ALP) which can quickly provide interpretable graphical summaries of distributional differences at any location x in the feature space. Our validation procedures scale to high dimensions and can potentially adapt to any type of data at hand. We demonstrate the effectiveness of LCT and ALP through a simulated experiment and applications to prediction and parameter inference for image data.
Many studies have reported associations between later-life cognition and socioeconomic position in childhood, young adulthood, and mid-life. However, the vast majority of these studies are unable to quantify how these associations vary over time and
with respect to several demographic factors. Varying coefficient (VC) models, which treat the covariate effects in a linear model as nonparametric functions of additional effect modifiers, offer an appealing way to overcome these limitations. Unfortunately, state-of-the-art VC modeling methods require computationally prohibitive parameter tuning or make restrictive assumptions about the functional form of the covariate effects. In response, we propose VCBART, which estimates the covariate effects in a VC model using Bayesian Additive Regression Trees. With simple default hyperparameter settings, VCBART outperforms existing methods in terms of covariate effect estimation and prediction. Using VCBART, we predict the cognitive trajectories of 4,167 subjects from the Health and Retirement Study using multiple measures of socioeconomic position and physical health. We find that socioeconomic position in childhood and young adulthood have small effects that do not vary with age. In contrast, the effects of measures of mid-life physical health tend to vary with respect to age, race, and marital status. An R package implementing VC-BART is available at https://github.com/skdeshpande91/VCBART
Spatial generalized linear mixed models (SGLMMs) are popular and flexible models for non-Gaussian spatial data. They are useful for spatial interpolations as well as for fitting regression models that account for spatial dependence, and are commonly
used in many disciplines such as epidemiology, atmospheric science, and sociology. Inference for SGLMMs is typically carried out under the Bayesian framework at least in part because computational issues make maximum likelihood estimation challenging, especially when high-dimensional spatial data are involved. Here we provide a computationally efficient projection-based maximum likelihood approach and two computationally efficient algorithms for routinely fitting SGLMMs. The two algorithms proposed are both variants of expectation maximization (EM) algorithm, using either Markov chain Monte Carlo or a Laplace approximation for the conditional expectation. Our methodology is general and applies to both discrete-domain (Gaussian Markov random field) as well as continuous-domain (Gaussian process) spatial models. Our methods are also able to adjust for spatial confounding issues that often lead to problems with interpreting regression coefficients. We show, via simulation and real data applications, that our methods perform well both in terms of parameter estimation as well as prediction. Crucially, our methodology is computationally efficient and scales well with the size of the data and is applicable to problems where maximum likelihood estimation was previously infeasible.
We study high-dimensional Bayesian linear regression with a general beta prime distribution for the scale parameter. Under the assumption of sparsity, we show that appropriate selection of the hyperparameters in the beta prime prior leads to the (nea
r) minimax posterior contraction rate when $p gg n$. For finite samples, we propose a data-adaptive method for estimating the hyperparameters based on marginal maximum likelihood (MML). This enables our prior to adapt to both sparse and dense settings, and under our proposed empirical Bayes procedure, the MML estimates are never at risk of collapsing to zero. We derive efficient Monte Carlo EM and variational EM algorithms for implementing our model, which are available in the R package NormalBetaPrime. Simulations and analysis of a gene expression data set illustrate our models self-adaptivity to varying levels of sparsity and signal strengths.
سجل دخول لتتمكن من نشر تعليقات
التعليقات
جاري جلب التعليقات
سجل دخول لتتمكن من متابعة معايير البحث التي قمت باختيارها
