Some models are useful, but how do we know which ones? Towards a unified Bayesian model taxonomy

Abstract

Probabilistic (Bayesian) modeling has experienced a surge of applications in almost all quantitative sciences and industrial areas. This development is driven by a combination of several factors, including better probabilistic estimation algorithms, flexible software, increased computing power, and a growing awareness of the benefits of probabilistic learning. However, a principled Bayesian model building workflow is far from complete and many challenges remain. To aid future research and applications of a principled Bayesian workflow, we ask and provide answers for what we perceive as two fundamental questions of Bayesian modeling, namely (a) "What actually is a Bayesian model?" and (b) "What makes a good Bayesian model?". As an answer to the first question, we propose the PAD model taxonomy that defines four basic kinds of Bayesian models, each representing some combination of the assumed joint distribution of all (known or unknown) variables (P), a posterior approximator (A), and training data (D). As an answer to the second question, we propose ten utility dimensions according to which we can evaluate Bayesian models holistically, namely, (1) causal consistency, (2) parameter recoverability, (3) predictive performance, (4) fairness, (5) structural faithfulness, (6) parsimony, (7) interpretability, (8) convergence, (9) estimation speed, and (10) robustness. Further, we propose two example utility decision trees that describe hierarchies and trade-offs between utilities depending on the inferential goals that drive model building and testing.

References in corpus (32)

• Towards A Rigorous Science of Interpretable Machine Learning
• To Explain or to Predict?
• Asymptotic Equivalence of Bayes Cross Validation and Widely Applicable Information Criterion in Singular Learning Theory
• Expectation Propagation for approximate Bayesian inference
• A Survey on the Explainability of Supervised Machine Learning
• Sparsity information and regularization in the horseshoe and other shrinkage priors
• The prior can generally only be understood in the context of the likelihood
• Markov Chain Monte Carlo: Can We Trust the Third Significant Figure?
• Misspecification in infinite-dimensional Bayesian statistics
• Flexible statistical inference for mechanistic models of neural dynamics
• Automatic Posterior Transformation for Likelihood-Free Inference
• CausalGAN: Learning Causal Implicit Generative Models with Adversarial Training
• Simulation Intelligence: Towards a New Generation of Scientific Methods
• Model Selection in Bayesian Neural Networks via Horseshoe Priors
• Graphical Test for Discrete Uniformity and its Applications in Goodness of Fit Evaluation and Multiple Sample Comparison
• Exploring the Nonlinear Cloud and Rain Equation
• Truncated proposals for scalable and hassle-free simulation-based inference
• Intuitive Joint Priors for Bayesian Linear Multilevel Models: The R2D2M2 prior
• An Evolutionary Approach to Dynamic Introduction of Tasks in Large-scale Multitask Learning Systems
• Bayesian Model Selection, the Marginal Likelihood, and Generalization
• A fully Bayesian sparse polynomial chaos expansion approach with joint priors on the coefficients and global selection of terms
• Detecting Model Misspecification in Amortized Bayesian Inference with Neural Networks
• JAGS, NIMBLE, Stan: a detailed comparison among Bayesian MCMC software
• Robust Neural Posterior Estimation and Statistical Model Criticism
• Implicit Deep Adaptive Design: Policy-Based Experimental Design without Likelihoods
• Sequential Neural Posterior and Likelihood Approximation
• JANA: Jointly Amortized Neural Approximation of Complex Bayesian Models
• Likelihood-Free Inference with Generative Neural Networks via Scoring Rule Minimization
• A Framework for Improving the Reliability of Black-box Variational Inference
• Physics-informed dynamic mode decomposition (piDMD)
• Amortized Bayesian Inference for Models of Cognition
• Locally Adaptive Shrinkage Priors for Trends and Breaks in Count Time Series

Cited by in corpus (2)

• Uncertainty Quantification and Propagation in Surrogate-based Bayesian Inference
• Gaussian distributional structural equation models: A framework for modeling latent heteroscedasticity