GP-BART improves BART's predictive performance by incorporating Gaussian process priors.
problem Lack of smoothness and explicit covariance structure in BART.
method GP-BART extends BART with Gaussian process priors for tree predictions.
result GP-BART outperforms traditional models in various applications.
This study converts BART to Gaussian process regression, revealing its limitations and potential improvements.
problem Understanding the Gaussian process limit of BART and its implications.
method Deriving and computing BART's prior covariance function, implementing the infinite trees limit as GP regression, and tuning hyperparameters.
result The Gaussian process limit of BART is inferior to standard BART but can be made competitive with proper hyperparameter tuning.
BART and MOTR-BART improve tree-based predictions with local linear models.
problem Non-linearity and high-order interactions in data.
method Bayesian Additive Regression Trees (BART) and Model Trees BART (MOTR-BART) using piecewise linear functions.
result MOTR-BART achieves equal or better performance with fewer trees than BART.
Oblique BART improves tree-based predictions.
problem Axis-aligned decision rules in BART can be suboptimal.
method Developed an oblique version of BART using data-adaptive hyperplane partitions.
result Oblique BART outperformed axis-aligned BART and other tree methods on benchmarks.
ASBART accelerates Soft BART for faster Bayesian regression.
problem Slow computation in Soft BART.
method Proposed ASBART, a variant of Soft BART.
result ASBART is about 10 times faster than Soft BART with similar accuracy.
GPU-accelerated BART speeds up Bayesian regression.
problem Long running time of BART makes it impractical for large datasets.
method GPU-enabled implementation of BART.
result BART is now 200x faster on GPUs.
Probit Monotone BART estimates binary outcomes using monotonic functions.
problem Estimating conditional mean functions for binary outcomes with monotonicity constraints.
method Proposes a new BART variant that incorporates monotonicity constraints for binary outcomes.
result Allows for more precise estimation of monotonic functions in binary outcome models.
New algorithm broadens BART models applicability.
problem Limited applicability of Bayesian additive regression trees (BART) models due to conditional conjugacy.
method Introduces a reversible jump Markov chain Monte Carlo algorithm for generalized BART models.
result Extends BART models to arbitrary generalized BART models without conditional conjugacy.
Bayesian Additive Regression Trees (BART) is a fully Bayesian approach to modeling with ensembles of trees. BART can uncover complex regression functions with high dimensional regressors in a fairly automatic way and provide Bayesian quantification of the uncertainty through the posterior. However, BART assumes IID nor…
BARK optimizes black-box functions using Bayesian Additive Regression Trees.
problem Bayesian optimization of complex, black-box functions with uncertainty quantification.
method BART Kernel using tree agreement for posterior over piecewise-constant functions, explored using MCMC.
result BARK obtains samples of Gaussian processes for function distributions, enabling acquisition functions for optimization.
BART is extended to handle various response variables.
problem Modeling nonlinear regression functions for diverse response types.
method Generalized Bayesian Additive Regression Trees (GBART) for exponential family distributions.
result The posterior concentrates at a minimax rate for certain response distributions.
SoftBart improves BART for high-noise modeling in science.
problem High noise in scientific data.
method Soft BART algorithm for Bayesian additive regression trees.
result Improves predictive performance and facilitates larger model integration.
flexBART improves BART for categorical predictors by creating flexible tree partitions.
problem Limitation of BART in handling categorical predictors with one-hot encoding.
method flexBART re-implements BART with regression trees that can assign multiple levels to both branches of a decision tree node, and proposes a new decision rule prior for spatial data.
result flexBART often yields improved predictive performance and scales better to larger datasets than existing BART implementations.
Infinite BART model selects number of trees and allows different functions for clusters.
problem Regression and classification analysis with automatic tree selection and cluster-specific functions.
method Incorporates an Indian Buffet process prior to select a subset of decision trees for each observation.
result Infinite BART model outperforms classic BART on simulated and real datasets.
We present a new package in R implementing Bayesian additive regression trees (BART). The package introduces many new features for data analysis using BART such as variable selection, interaction detection, model diagnostic plots, incorporation of missing data and the ability to save trees for future prediction. It is …
Proposes extensions to semi-parametric models using BART for shared covariates.
problem Avoiding poor coverage properties and reducing bias in linear predictor estimates.
method Modifies tree-generation moves in BART to handle shared covariates between linear and non-parametric components.
result Competitive performance in modelling complex interactions and predicting student achievements.
Bayesian Additive Distribution Regression (DistBART) predicts distributions from grouped data.
problem Predicting distributions from grouped data with varying characteristics.
method Bayesian nonparametric approach using BART for modeling the regression function.
result Empirical and theoretical evidence supports DistBART's effectiveness in learning from low-dimensional marginals.
Proposes using external data to improve predictions in medical applications with limited samples.
problem Small sample sizes and complex covariate-response relationships in medical data.
method Integrates external co-data into Bayesian Additive Regression Trees (BART) using an empirical Bayes framework.
result Improves prediction accuracy compared to standard BART, especially for nonlinear relationships.
Additive regression trees are flexible non-parametric models and popular off-the-shelf tools for real-world non-linear regression. In application domains, such as bioinformatics, where there is also demand for probabilistic predictions with measures of uncertainty, the Bayesian additive regression trees (BART) model, i…
We develop a Bayesian "sum-of-trees" model where each tree is constrained by a regularization prior to be a weak learner, and fitting and inference are accomplished via an iterative Bayesian backfitting MCMC algorithm that generates samples from a posterior. Effectively, BART is a nonparametric Bayesian regression appr…
Functional BART adds shape priors to Bayesian tree regression for better curve fitting.
problem Regression with function-on-scalar data and shape constraints.
method Bayesian tree structure with spline representations, customized Bayesian backfitting algorithm, shape priors.
result Improved estimation and prediction accuracy with shape priors.
Bayesian Probabilistic Integration uses BART for high-dimensional, non-smooth functions.
problem Bayesian quadrature's limitations in high-dimensional or non-smooth functions.
method Bayesian Additive Regression Trees (BART) priors for numerical integration.
result Explicit convergence rates can be obtained in various settings.
Bayesian additive regression trees (BART) (Chipman et. al., 2010) is a powerful predictive model that often outperforms alternative models at out-of-sample prediction. BART is especially well-suited to settings with unstructured predictor variables and substantial sources of unmeasured variation as is typical in the so…
BART improves predictive performance but slows down with more data.
problem Understanding and improving the computational efficiency of BART with large datasets.
method Asymptotic analysis of a modified BART sampler, focusing on hitting time of high posterior density sets.
result The convergence time of the BART sampler increases with the number of training samples due to multi-modality, but can be mitigated by increasing the number of trees or raising the sampler temperature.
SBT model uses randomized sharding and sub-models to improve Bayesian Additive Regression Trees.
problem Improving efficiency and accuracy of Bayesian Additive Regression Trees.
method Randomized sharding, sub-models, intersection tree structure, optimal design.
result Theoretical optimal weights and worst-case complexity of SBT model.
Bayesian approach for estimating heterogeneous treatment effects in RDD designs.
problem Heterogeneity in treatment effects in RDD designs can lead to misleading conclusions.
method Direct Bayesian Additive Regression Trees (BART) for modeling heterogeneous treatment effects.
result Flexibly captures complicated structures of heterogeneous treatment effects as a function of covariates.
This article proposes Multinomial Probit Bayesian Additive Regression Trees (MPBART) as a multinomial probit extension of BART - Bayesian Additive Regression Trees (Chipman et al (2010)). MPBART is flexible to allow inclusion of predictors that describe the observed units as well as the available choice alternatives. T…
Lower bound on BART's mixing time increases with data points.
problem Slow mixing time in BART's MCMC chains.
method Simplified BART with a single tree and reduced MCMC moves.
result Mixing time grows exponentially with data points.
BAVART model combines VAR and BART for non-linear forecasting.
problem Overly restrictive linearity assumption in VAR models.
method Combining VAR with Bayesian additive regression trees (BART).
result BAVART model yields highly competitive forecasts.
We present a method for incorporating missing data in non-parametric statistical learning without the need for imputation. We focus on a tree-based method, Bayesian Additive Regression Trees (BART), enhanced with "Missingness Incorporated in Attributes," an approach recently proposed incorporating missingness into deci…
Two methods use BART to model missing data in leaf photosynthetic trait data.
problem Handling missing data in multivariate outcomes with non-ignorable mechanisms.
method Bayesian Additive Regression Trees (BART) for joint modeling of data and missingness indicators.
result Both methods effectively recover various missingness mechanisms and outperform existing approaches.
BFTS uses Bayesian Additive Regression Trees for improved personalized mobile health interventions.
problem Adapting to complex, non-linear user behaviors in personalized mobile health interventions.
method Bayesian Forest Thompson Sampling (BFTS) integrates Bayesian Additive Regression Trees (BART) into the exploration loop of contextual bandits.
result BFTS achieves state-of-the-art regret on tabular benchmarks and improves engagement rates by over 30% in a behavioral intervention study.
Develops a new multivariate regression model for complex outcomes.
problem Flexible, heterogeneous, and residual-dependent multivariate regression problems.
method MultiVCBART framework with Graphical Horseshoe priors.
result Empirically outperforms existing models on sparse, high-dimensional datasets.
Bayesian model estimates treatment effects near cutoffs in regression discontinuity designs.
problem Estimating conditional average treatment effects in regression discontinuity designs.
method Develops a Bayesian additive regression tree (BART) model with linear leaf-level regressions.
result Adapts to different slopes on the running variable near the cutoff, providing interpretable inference.
Combines BART and Gaussian process for spatial covariate prediction with uncertainty.
problem Improving spatial prediction models with nonlinear and interaction covariates.
method Bayesian Additive Regression Trees (BART) combined with Gaussian process for spatial dependence.
result Effective in reducing computational burden through INLA and MCMC.
Bayesian Additive Regression Networks use neural networks for regression tasks.
problem Regression tasks with small neural networks and ensemble learning.
method Bayesian Additive Regression Tree principles applied to small neural networks, Gibbs sampling for ensemble learning.
result BARN provides more consistent and often more accurate results than shallow neural networks, BART, and ordinary least squares.
The paper introduces a new model to correct bias in treatment effect estimates due to sample selection.
problem Bias in treatment effect estimates due to sample selection.
method Type 2 Tobit Bayesian Additive Regression Trees (TOBART-2) with Dirichlet Process Mixture distribution and soft trees.
result Corrects bias in treatment effect estimates by accounting for nonlinearities and model uncertainty.
Bayesian nonparametric machine learning improves instrumental variable inference.
problem Estimating causal effects with nonlinear relationships.
method Bayesian Additive Regression Trees (BART) for estimating functions and Dirichlet Process mixtures for error terms.
result Dramatic improvements in inference with nonlinear data, no manual tuning required.
Paper proposes a new method to improve BART model predictions outside training data range.
problem Improving prediction and prediction intervals for BART models at extrapolation points.
method Gaussian processes are added to BART leaf nodes for extrapolation.
result The new method outperforms standard BART and frequentist resampling methods in simulations.
This paper develops a novel stochastic tree ensemble method for nonlinear regression, which we refer to as XBART, short for Accelerated Bayesian Additive Regression Trees. By combining regularization and stochastic search strategies from Bayesian modeling with computationally efficient techniques from recursive partiti…
A Bayesian approach to multilabel classification using tree-based models.
problem Challenges in multilabel classification due to complex label relationships and correlations.
method Bayesian Additive Regression Trees (BART) framework for modeling multilabel classification.
result Improved predictive performance compared to other models, including an oracle model.
Proposes K-Fold Causal BART for improved CATE estimation.
problem Improving estimation of Conditional Average Treatment Effects (CATE).
method K-Fold Causal Bayesian Additive Regression Trees (K-Fold Causal BART).
result K-Fold Causal BART is not state-of-the-art for ATE and CATE estimation in the IHDP dataset, but provides insights into model robustness and evaluation methods.
Using ensemble methods for regression has been a large success in obtaining high-accuracy prediction. Examples are Bagging, Random forest, Boosting, BART (Bayesian additive regression tree), and their variants. In this paper, we propose a new perspective named variable grouping to enhance the predictive performance. Th…
Developed a flexible Bayesian g-formula for causal survival analysis with time-dependent confounding.
problem Estimating causal survival curves in longitudinal observational studies with time-varying treatments and confounding.
method Incorporated Bayesian Additive Regression Trees (BART) into the g-formula to model time-evolving generative components and mitigate bias due to model misspecification.
result Demonstrated improved empirical performance and practical utility of the proposed method through simulations and real-world data analysis.
Proposes a new model to handle noisy data in scientific research.
problem Measurement error in noisy data settings.
method Measurement error BART (meBART) integrates measurement error in Bayesian additive regression trees.
result meBART provides more accurate parameter estimation, robust uncertainty quantification, and superior predictive performance.
This work improves mixing rates for Bayesian CART, a key component of BART.
problem Understanding and improving mixing rates for Bayesian inference with MCMC.
method Derived upper bounds on mixing times, provided sufficient conditions for polynomial mixing, and proposed Twiggy Bayesian CART.
result Twiggy Bayesian CART achieves polynomial mixing without assuming signal connectivity.
The paper studies causal effects of multiple treatments in healthcare databases with rare outcomes.
problem Estimating causal effects of multiple treatments in healthcare databases with rare outcomes.
method The paper designs three sets of simulations and compares the operating characteristics of three types of methods: Bayesian Additive Regression Trees (BART), regression adjustment on multivariate spline of generalized propensity scores (RAMS), and inverse probability of treatment weighting (IPTW) with multinomial logistic regression or generalized boosted models.
result BART and RAMS provide lower bias and mean squared error compared to IPTW methods.
SBAMDT uses adaptive soft splits to model complex decision boundaries.
problem Limited ability of standard decision trees to capture complex decision boundaries.
method Probabilistic additive decision tree model with adaptive soft multivariate splits.
result Demonstrated improved predictive performance on synthetic and real datasets.