Home > OS >  How to shape the labels to match the output layer for classification(ANN)?
How to shape the labels to match the output layer for classification(ANN)?

Time:05-14

I am trying to learn and understand how to implement multiclass classification using ANN. In my case, I have 16 classes(0-15), and my label dataset contains one column with the label values. So I know that the output layer should have the same number of neurons as the classes. When I created the output layer with 16 neurons, I got the following error message:

Shapes (32, 1) and (32, 16) are incompatible

I followed the solutions from similar questions:

y_train= tf.one_hot(y_train, 16)

and I got the following error:

ValueError: Shapes (32, 1, 16) and (32, 16) are incompatible

I understand that the problem is with the shape of my labels, but I have no idea how to fix it.

I appreciate any help you can provide.

CodePudding user response:

The solution was to use the following:

from tensorflow.keras.utils import to_categorical    
y_cat_test = to_categorical(y_test,16)
  • Related