We need a method to define a population model with uncertainty. This can be in an abstract sense or in a more specific sense. For example, just marginalizing over the class prior could be very useful, and would be much more tractable than marginalizing hyperparameters related to the shape of the population predictions in the observable parameter space.
We need a method to define a population model with uncertainty. This can be in an abstract sense or in a more specific sense. For example, just marginalizing over the class prior could be very useful, and would be much more tractable than marginalizing hyperparameters related to the shape of the population predictions in the observable parameter space.