[SOLVED] how to implement hamming loss as a custom metric in keras model

how to implement hamming loss as a custom metric in keras model

how to implement hamming loss as a custom metric in keras model
I have a multilabel classification with 6 classes

model.compile(loss='binary_crossentropy', optimizer='adam', metrics=['accuracy',hamming_loss])

I tried using

from sklearn.metrics import hamming_loss
def custom_hl(y_true, y_pred):    
  return hamming_loss(y_true, y_pred)

which doesn't work because i have the y_true , y_pred as follow

YTRUE
Tensor("Cast_10:0", shape=(None, 6), dtype=float32)
YPRED
Tensor("model_1/dense_1/Sigmoid:0", shape=(None, 6), dtype=float32)

also tried the function in this question and it doesn't work Getting the accuracy for multi-label prediction in scikit-learn is there any way I can get the hamming loss as metric in keras thanks for any help

Solution

so i found a way and

def Custom_Hamming_Loss(y_true, y_pred):
  return K.mean(y_true*(1-y_pred)+(1-y_true)*y_pred)

def Custom_Hamming_Loss1(y_true, y_pred):
  tmp = K.abs(y_true-y_pred)
  return K.mean(K.cast(K.greater(tmp,0.5),dtype=float))

source:https://groups.google.com/g/keras-users/c/_sjndHbejTY?pli=1