Loading Events

Laura Ventura (University of Padua)

29 April 2016 @ 12:00

 

  • Past event

Details

Date:
29 April 2016
Time:
12:00
Event Category:

Robust Approximate Bayesian Inference

The likelihood function is the basis of both frequentist and Bayesian methods. However, the stability of likelihood-based procedures requires strict adherence to the model assumptions: mild deviations from the model can lead to misleading inferential results. A possible Bayesian solution to robustness is to use a of robust pseudo likelihood, such as the quasi- and the empirical likelihoods, in the Bayes formula. However, in multiparameter problems the quasi-likelihood is cumbersome, while with small samples the empirical likelihood can be unstable. Moreover, some estimating functions can be difficult to evaluate, which rules out the use of the empirical likelihood and other existing methods (Bayesian bootstrap, Bayesian GMM, etc.). To avoid these drawbacks, in this contribution we focus on posterior distributions from robust unbiased M-estimating functions. These robust posteriors are obtained via Approximate Bayesian Computation (ABC) with the estimating functions used as summary statistics. Theory and examples with robust M-estimating functions are discussed.