I have a csv files which contain pixel neighboorhood information. Here an example of the dataset:

0 0 1.875223e+01
1 0 1.875223e+01
2 0 2.637685e+01
3 0 2.637685e+01
4 0 2.637685e+01
5 0 2.637685e+01
6 0 2.637685e+01
7 0 2.637685e+01
8 0 2.637685e+01
9 0 1.875223e+01

I would like to know if it is a good idea to apply a normalization on this dataset before training the convolutional autoencoder ?

Normalize the value data

scaler = StandardScaler()
values_scaled = scaler.fit_transform(values.reshape(-1, 1))

My second question, should I use MSE or binary_crossentropy as loss function?


1 Answer 1


Yes, it is a good idea to normalize the data. You can also consider using the MinMax scaler, as it might improve the performance. For your second question, most people use binary_crossentropy, however it also has some cons. There is a good discussion here.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.