search

rmattson1008 / ornet

Organellar segmentation, tracking, and network modeling.
MIT License
0 stars 0 forks source link

Fold wild-type into train set #21

Closed rmattson1008 closed 1 year ago

rmattson1008 commented 1 year ago

So far have just been separating hyperfused and hypofused with the model, not touching the wild-type. Now that I've shown the model can represent these two I want to fold some in, randomly to both original classes. Conceptually, this might teach the model that fusion and fission can occur in balance. After all, the lofty goal is to identify fusion/fission processes, not LLO and MDIVI. At worst, this would add some regularization to the model and make it more robust when it sees new videos.

rmattson1008 commented 1 year ago

change made, need to test in actual program