Apply in multi-label task

[oguriq]

Hi, I want to know if this function can be directly applied in the multi-label task. I used to use the binary-crossentropy as my loss function in the multi-label task , now I want to use focal loss to replace it. Should I make some changes or not ? Anybody help?

[kmh93]

I think you can not use it directly

[umbertogriffo]

@oguriq take a look at https://github.com/umbertogriffo/focal-loss-keras

[heya5]

When I used softmax for binary classification, the program ran without error, but I deduced that the calculation process was wrong. We can see that only 0 and 1 are compared in this code, so it can't be used for multi-classification.