4 ms·
"Overfitting" would be a little more specific. Basically the task was easier to accomplish by memorizing the data than by generalizing. This was pretty clear f
by soraki_soladead 8y ago
"Overfitting" would be a little more specific. Basically the task was easier to accomplish by memorizing the data than by generalizing.
This was pretty clear from the original CycleGAN paper[0] (figure 4; bottom row) since it was generating trees in the exact positions of the source image where the map had no indication of tree placement.
0. https://arxiv.org/pdf/1703.10593v1.pdf https://arxiv.org/pdf/1703.10593v1.pdf
- phyalow 8y agoSo in anycase the model had too many degrees of freedom, was not parsimoinous and therefore had a low information criterion. If your model can encode the entire information set, then you arent prediciting you are simply recalling. What a silly article.
- candiodari 8y agoNot at all. A CycleGAN is a trick where the neural network gets to ask another network questions, and both networks get judged on the answers. The first network found a trick to very subtly embed the perfect answer in the question, and therefore, scored very well without solving the actual problem.
- ashelmire 8y agoI don’t think this is overfitting. It’s really a form of leakage by the design; it seems that the loss function is a result of the outcome of both networks (cyclegan is really two gans connected it seems), which means the loss gets shared and backpropagated through both networks. I think the architecture is bad. Maybe you could stop the network from embedding extra data with some sort of normalization or max pooling at the very end of the network.
- soraki_soladead 8y agoI believe it's both. The produced images do not encode pixel-perfect details necessary for reconstruction. These details are reconstructed from the imperceptible differences in the outputs which act as markers _and_ the model's memorization of the content at those locations.
- ashelmire 8y agoAh, that's a good point.