Hi,
Is there any best practice about how to implement an auto-encoder based on resnet architecture ? i.e. how to decrease in the dimensionality with keeping skip connections safe from the bottleneck ?
Best
Hi,
Is there any best practice about how to implement an auto-encoder based on resnet architecture ? i.e. how to decrease in the dimensionality with keeping skip connections safe from the bottleneck ?
Best
Ahmed, if the purpose of the auto-encoder is feature extraction, here’s what could potentially be interesting: