2

1

I have been reading a paper by Burda et al. on Importance Weighted Autoencoders(IWAE) but I can't quite grasp what they mean by sampling the terms **h**1...**h**k. Do they mean you have separate models from which you sample and then average over?

2

1

I have been reading a paper by Burda et al. on Importance Weighted Autoencoders(IWAE) but I can't quite grasp what they mean by sampling the terms **h**1...**h**k. Do they mean you have separate models from which you sample and then average over?

post something more from the paper explaining what are those terms – Francesco Pegoraro – 2018-10-03T12:58:35.217