make ensemble predictions simple, in an effort to produce more conservative posterior approimations

[bkmi]

This paper by Gilles and co, Averting A Crisis In Simulation-Based Inference, advocates using ensembles to combat non-conservative posterior estimates.

I think it makes sense to allow for simple ensemble averaging of predictions in swyft.

[cweniger]

• Is parallel training of estimators in the same network enough?
• Could this be achieved with MC dropout?

[bkmi]

I think a bagging approach would likely be best.