What should I do if I have 11 different labels (each column has 0 or 1 values, but for one row more than one column can have “1” value)? What should my last layer look like? What should my loss function be?
Hi @Lusia, This will be a multi-label classification task as for some rows we have more than one 1 per column. you can use a dense layer with 11 neurons (1 neuron per each class) and can use the activation as sigmoid.
tf.keras.layers.Dense(10,activation='sigmoid')
Thank You.