I'm working on a classification project and need to do one hot encoding on my data set. I'm just wondering what is the difference between tf.keras.backend.one_hot and keras.utils.to_categorical, and is one of them preferred over the other?
$\begingroup$
$\endgroup$
Add a comment
|
1 Answer
$\begingroup$
$\endgroup$
The difference is the input and output.
The utils.to_categorical function takes a vector as input and returns a matrix with one hot encoded rows.
The backend.one_hot can be called as a transformation step inbetween a compute sequence of tensors.
It takes nD tensor as input and returns n+1 D one hot encoded tensor.
