weighted_categorical_crossentropy

gistfile1.txt

def weighted_categorical_crossentropy(weights):

    weights = K.variable(weights)
        
    def loss(y_true, y_pred):
        # scale predictions so that the class probas of each sample sum to 1
        y_pred /= K.sum(y_pred, axis=-1, keepdims=True)
        # clip to prevent NaN's and Inf's
        y_pred = K.clip(y_pred, K.epsilon(), 1 - K.epsilon())
        # calc
        lss = y_true * K.log(y_pred) * weights
        lss += (1 - y_true) * K.log(1 - y_pred) * weights        
        lss = -K.sum(lss, -1)
        return lss
    
    return loss
    
weights = np.array([0.5,5,10]) 
loss = weighted_categorical_crossentropy(weights)