Open saddy001 opened 8 years ago
It would be very nice to have the ROC AUC as classifier loss. Theanets has hidden the tensor logic from me so successfully, that I'm not able to write the loss function on my own.
Someone has already done the work for pylearn2.
It would be very nice to have the ROC AUC as classifier loss. Theanets has hidden the tensor logic from me so successfully, that I'm not able to write the loss function on my own.