view article

Figure 7
Flow diagram showing a variational Bayesian inference scheme using a differentiable simulation. The posterior distribution is approximated by a multivariate Gaussian qθ(x) with parameters θ. Samples are taken from this distribution and used as the inputs for a simulation. The simulation output and measured intensity profiles are used to calculate the ELBO where p(x, y) = p(y|x)p(x). Using automatic differentiation, the gradient of the ELBO with respect to θ is calculated. By minimizing the ELBO, the difference between qθ and p(x|y) is also minimized.

Journal logoJOURNAL OF
SYNCHROTRON
RADIATION
ISSN: 1600-5775
Follow J. Synchrotron Rad.
Sign up for e-alerts
Follow J. Synchrotron Rad. on Twitter
Follow us on facebook
Sign up for RSS feeds