4 ms·
This is somewhat tangental, but what do you think about Ladder Networks? (https://arxiv.org/abs/1507.02672 https://arxiv.org/abs/1507.02672, https://arxiv.org/
by ma2rten 10y ago
This is somewhat tangental, but what do you think about Ladder Networks?
(https://arxiv.org/abs/1507.02672 https://arxiv.org/abs/1507.02672, https://arxiv.org/abs/1511.06430 https://arxiv.org/abs/1511.06430)
I was really impressed by their results, but I haven't seen it being applied successfully to other applications.
- jostmey 10y agoDisclaimer, I haven't looked at their work to closely. I think the process they use to select the hyperparameters overfits on the labels in the validation set. The true size of the labelled data includes the 100 from the training set and all labels in the validation set.