One Hot Encoding list of strings

Question

I have a list of strings which serve as labels for my classification problem (image recognition with a Convolutional Neural Network). These labels consist of 5-8 characters (numbers from 0 to 9 and letters from A to Z). To train my neural network I would like to one hot encode the labels. I wrote a code to encode one label but I am still experiencing difficulties when trying to apply the code to a list.

Here is my code for one label which works fine:

from numpy import argmax
# define input string
data = '7C24698'
print(data)
# define universe of possible input values
characters = '0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZ '
# define a mapping of chars to integers
char_to_int = dict((c, i) for i, c in enumerate(characters))
int_to_char = dict((i, c) for i, c in enumerate(characters))
# integer encode input data
integer_encoded = [char_to_int[char] for char in data]
print(integer_encoded)
# one hot encode
onehot_encoded = list()
for value in integer_encoded:
    character = [0 for _ in range(len(characters))]
    character[value] = 1
    onehot_encoded.append(character)
print(onehot_encoded)
# invert encoding
inverted = int_to_char[argmax(onehot_encoded[0])]
print(inverted)

I now want to get the same output for list of labels and store the output in a new list:

list_of_labels = ['7C24698', 'NDK745']
encoded_labels = []

How can I do this?

Answer 1

You can use LabelBinarizer from scikit-learn:

from sklearn.preprocessing import LabelBinarizer

>>> labels = ["first", "second", "third"]
>>> lb = LabelBinarizer()
>>> lb.fit(labels)
>>> lb.transform(labels)
array([[1, 0, 0],
       [0, 1, 0],
       [0, 0, 1]])

And to convert back the one-hot encoded labels back to string values:

>>> encoded_labels = [
  [1, 0, 0],
  [0, 1, 0],
  [0, 0, 1]
]
>>> lb.inverse_transform(encoded_labels)
array(['first', 'second', 'third'])