Credal model averaging for classification: representing prior ignorance and expert opinions

DOI10.1016/J.IJAR.2014.07.001zbMATH Open1388.68263arXiv1405.3559OpenAlexW2171871229MaRDI QIDQ473394FDOQ473394

Authors: Andrea Mignatti, Giorgio Corani 

Publication date: 24 November 2014

Published in: International Journal of Approximate Reasoning (Search for Journal in Brave)

Abstract: Bayesian model averaging (BMA) is the state of the art approach for overcoming model uncertainty. Yet, especially on small data sets, the results yielded by BMA might be sensitive to the prior over the models. Credal Model Averaging (CMA) addresses this problem by substituting the single prior over the models by a set of priors (credal set). Such approach solves the problem of how to choose the prior over the models and automates sensitivity analysis. We discuss various CMA algorithms for building an ensemble of logistic regressors characterized by different sets of covariates. We show how CMA can be appropriately tuned to the case in which one is prior-ignorant and to the case in which instead domain knowledge is available. CMA detects prior-dependent instances, namely instances in which a different class is more probable depending on the prior over the models. On such instances CMA suspends the judgment, returning multiple classes. We thoroughly compare different BMA and CMA variants on a real case study, predicting presence of Alpine marmot burrows in an Alpine valley. We find that BMA is almost a random guesser on the instances recognized as prior-dependent by CMA.

Full work available at URL: https://arxiv.org/abs/1405.3559

Recommendations

• Credal ensembles of classifiers
• Bayesian model averaging: A tutorial. (with comments and a rejoinder).
• Tree-based credal networks for classification
• Bayesian Model Averaging for Linear Regression Models
• On the quantification of model uncertainty: a Bayesian perspective

zbMATH Keywords

logistic regressionBayesian model averagingimprecise probabilitiescredal classificationcredal model averaging

Mathematics Subject Classification ID

Bayesian inference (62F15) General nonlinear regression (62J02) Reasoning under uncertainty in the context of artificial intelligence (68T37)

Cites Work

• Bayesian model averaging: A tutorial. (with comments and a rejoinder).
• Model uncertainty
• An exploration of aspects of Bayesian multiple testing
• Title not available (Why is that?)
• Objective Bayesian model selection in Gaussian graphical models
• Decision making under uncertainty using imprecise probabilities
• Title not available (Why is that?)
• Evaluating credal classifiers by utility-discounted predictive accuracy
• Imprecise probability models for learning multinomial distributions from data. Applications to learning credal networks
• Learning reliable classifiers from small or incomplete data sets: the naive credal classifier 2

Cited In (3)

• Multilabel predictions with sets of probabilities: the Hamming and ranking loss cases
• Inference after checking multiple Bayesian models for data conflict and applications to mitigating the influence of rejected priors
• Credal ensembles of classifiers

This page was built for publication: Credal model averaging for classification: representing prior ignorance and expert opinions

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q473394)