search

AllenCellModeling / pytorch_integrated_cell

Integrated Cell project implemented in pytorch
43 stars 11 forks source link

Discussion: Loss function #77

Closed donovanr closed 3 years ago

donovanr commented 4 years ago

Issue summary

Talk about how loss function can affect results

Details

It would also be nice in the discussion to offer some thoughts and speculations about other ways to tailor the loss function for specific cell biological applications. Rory had some interesting suggestions along these lines when he spoke to us on Friday.

TODO

donovanr commented 4 years ago

@gregjohnso you have the most direct experience here -- maybe add some text around MSE vs BCE, how the choice affect results and interoperability? I personally don't want to go to hard on MSE -> gaussian stuff but anything around how different losses would be better or worse for different structure patterns, BCE good for segmentations, etc might be good to add.

gregjohnso commented 3 years ago

Added content in \subsubsection{Results changing with image normalization could be a big deal}