Preprocessing Encrypted Data#

USEncrypt.ai allows users to preprocess encrypted data before using them to train machine learning models. The following preprocessing functions are available:

clip_by_value

Clips the input to values in the specified range.

encode_one_hot

Encodes categorical features as a one-hot numeric array based on the unique values in each feature.

minmax_scale

Transforms the given input data using its minimum and maximum values.

normalize

Scales input vectors individually to the unit norm.

scale

Scales the input around a mean and unit standard deviation.

train_test_split

Splits the input arrays into training and test sets.

train_val_test_split

Splits the input arrays into training, validation, and test sets.