`

( events)   Timezone: »  
Workshop
Tue Dec 14 04:55 AM -- 04:30 PM (PST)
Your Model is Wrong: Robustness and misspecification in probabilistic modeling
Diana Cai · Sameer Deshpande · Michael Hughes · Tamara Broderick · Trevor Campbell · Nick Foti · Barbara Engelhardt · Sinead Williamson





Workshop Home Page

Probabilistic modeling is a foundation of modern data analysis -- due in part to the flexibility and interpretability of these methods -- and has been applied to numerous application domains, such as the biological sciences, social and political sciences, engineering, and health care. However, any probabilistic model relies on assumptions that are necessarily a simplification of complex real-life processes; thus, any such model is inevitably misspecified in practice. In addition, as data set sizes grow and probabilistic models become more complex, applying a probabilistic modeling analysis often relies on algorithmic approximations, such as approximate Bayesian inference, numerical approximations, or data summarization methods. Thus in many cases, approximations used for efficient computation lead to fitting a misspecified model by design (e.g., variational inference). Importantly, in some cases, this misspecification leads to useful model inferences, but in others it may lead to misleading and potentially harmful inferences that may then be used for important downstream tasks for, e.g., making scientific inferences or policy decisions.

The goal of the workshop is to bring together researchers focused on methods, applications, and theory to outline some of the core problems in specifying and applying probabilistic models in modern data contexts along with current state-of-the-art solutions. Participants will leave the workshop with (i) exposure to recent advances in the field, (ii) an idea of the current major challenges in the field, and (iii) an introduction to methods meeting these challenges. These goals will be accomplished through a series of invited and contributed talks, poster spotlights, poster sessions, as well as ample time for discussion and live Q&A.

Welcome remarks (Talk)
How to train your model when it's wrong: Bayesian nonparametric learning in M-open (Talk)
Invite Talk 1 Q&A (Q&A)
BayesCG: A probabilistic numeric linear solver (Talk)
Invited Talk 2 Q&A (Q&A)
Individual discussions in Gathertown (Gathertown discussion)
Bayesian Calibration of imperfect computer models using Physics-informed priors (Talk)
Invariant Priors for Bayesian Quadrature (Talk)
Poster Session I in Gathertown (Poster session)
Research panel (Discussion panel)
Individual discussions in Gathertown (Gathertown discussion)
Uncertainty estimation under model misspecification in neural network regression (Talk)
Your Bandit Model is Not Perfect: Introducing Robustness to Restless Bandits Enabled by Deep Reinforcement Learning (Talk)
Invited Talk 3 Q&A (Q&A)
Bayesian Model Averaging is not Model Combination: A PAC-Bayesian Analysis of Deep Ensembles (Talk)
Individual discussions in Gathertown (Gathertown discussion)
PAC^m-Bayes: Narrowing the Empirical Risk Gap in the Misspecified Bayesian Regime (Talk)
Bayesian Data Selection (Talk)
Statistically Robust Inference with Stochastic Gradient Algorithms (Talk)
Invited Talk 4 Q&A (Q&A)
Individual discussions in Gathertown (Gathertown discussion)
Your Model is Wrong (but Might Still Be Useful) (Talk)
Invited Talk 5 Q&A (Q&A)
Statistical and Computational Tradeoffs in Variational Bayes (Talk)
Invited Talk 6 Q&A (Q&A)
Poster session II in Gathertown + End (Poster session)
Invariant Priors for Bayesian Quadrature (Poster)
Blindness of score-based methods to isolated components and mixing proportions (Poster)
Inferior Clusterings in Misspecified Gaussian Mixture Models (Poster)
A shared parameter model accounting for drop-out not at random in a predictive model for systolic bloodpressure using the HUNT study (Poster)
Forcing a model to be correct for classification (Poster)
Relaxing the I.I.D. Assumption: Adaptively Minimax Optimal Regret via Root-Entropic Regularization (Poster)
Measuring the sensitivity of Gaussian processes to kernel choice (Poster)
Bounding Wasserstein distance with couplings (Poster)
Your Bandit Model is Not Perfect: Introducing Robustness to Restless Bandits Enabled by Deep Reinforcement Learning (Poster)
Composite Goodness-of-fit Tests with Kernels (Poster)
Make cross-validation Bayes again (Poster)
Robust Bayesian Inference for Simulator-based Models via the MMD Posterior Bootstrap (Poster)
PAC^m-Bayes: Narrowing the Empirical Risk Gap in the Misspecified Bayesian Regime (Poster)
Evaluating Bayesian Hierarchical Models for sc-RNA seq Data (Poster)
Fast approximate BayesBag model selection via Taylor expansions (Poster)
Uncertainty estimation under model misspecification in neural network regression (Poster)
On Counterfactual Analysis of Differentiable Functionals (Poster)
Robust Generalised Bayesian Inference for Intractable Likelihoods (Poster)
Bayesian Data Selection (Poster)
Diversity and Generalization in Neural Network Ensembles (Poster)
Boosting heterogeneous VAEs via multi-objective optimization (Poster)
Bayesian Calibration of imperfect computer models using Physics-informed priors (Poster)
Influential Observations in Bayesian Regression Tree Models (Poster)