Convert multi-labels of an instance to one-hot encoding.

one_hot_multi_label_encoding.py

import numpy as np

n_classes = 10
class_matrix = np.eye(n_classes)
"""
class_matrix should be obtained as below:
array([[1, 0, 0, 0, 0, 0, 0, 0, 0, 0],
       [0, 1, 0, 0, 0, 0, 0, 0, 0, 0],
       [0, 0, 1, 0, 0, 0, 0, 0, 0, 0],
       [0, 0, 0, 1, 0, 0, 0, 0, 0, 0],
       [0, 0, 0, 0, 1, 0, 0, 0, 0, 0],
       [0, 0, 0, 0, 0, 1, 0, 0, 0, 0],
       [0, 0, 0, 0, 0, 0, 1, 0, 0, 0],
       [0, 0, 0, 0, 0, 0, 0, 1, 0, 0],
       [0, 0, 0, 0, 0, 0, 0, 0, 1, 0],
       [0, 0, 0, 0, 0, 0, 0, 0, 0, 1]]) 
"""      
# for an instance with labels `y` = [0, 1, 3, 6]
# so, the new encoded labels must include the lines in 
# class_matrix with index [0, 1, 3, 6]. It becomes easy now.

y_encoded = class_matrix[y, :]
# We are done here.