Skip to main content
Research Highlight

A Novel Method for Bayesian Experimental Design with Implicit Models

The Science

ORNL researchers developed a stochastic approximate gradient ascent method to reduce posterior uncertainty in Bayesian experimental design involving implicit models.  

This method:

  • introduces a neural-network-based mutual information estimator into the objective function to replace the  computationally intractable likelihood function, and 
  • can simultaneously train the mutual information estimator and search for an optimal design. 

The Impact

This method:

  • achieves designs with significantly improved confidence, i.e., small uncertainty of the posterior distribution when compared to the state-of-the-art Bayesian optimization (BO) method, and
  • avoids intractable overhead cost of BO in solving high-dimensional problems. 

PI: Guannan Zhang
Publication: J. Zhang, S. Bi, and G. Zhang, A hybrid gradient method to designing Bayesian experiments for implicit models, NeurIPS Workshop Proceeding on Machine Learning and the Physical Sciences, Dec. 2020. (PDF file).
ASCR Program/Facility: DOE ASCR Applied Math 
Funding: DOE ASCR Applied Math (UQ4ML), and the ORNL AI Initiative