Location visible to members
Gaussian processes are a flexible way to specify a prior distribution over a continuous but unknown function that generates observable outcomes. Gaussian processes can also be seen as a continuous generalization of multilevel models where the parameters are allowed to vary from one discrete group to another. The statistical software Stan is well-suited for Bayesian estimation of models with Gaussian processes, but the user may have to overcome a variety of challenges, including specifying priors on the hyperparameters to a Gaussian process, interpreting the results, and obtaining results within a reasonable amount of time in moderately-sized datasets. This talk will provide strategies for overcoming these challenges.
Ben Goodrich from Columbia University is Stan Core Developer and in particular the main developer of the rstanarm R package (together with Jonah Gabry).
With registering and participating in the event you agree to be photographed by other attendees. The use case of these photos is to share, in social media and beyond, impressions from the event.
These photos can be used by the attendees and their employers for such purposes. You will be informed in detail prior to the meetup start and will have the opportunity to leave in case you object.