使用Keras np_utils.to_categorical的问题

Question

我正在尝试将一个单一的整数向量数组生成一个单热矢量数组，keras将能够使用它来适应我的模型。这是代码的相关部分：

Y_train = np.hstack(np.asarray(dataframe.output_vector)).reshape(len(dataframe),len(output_cols))
dummy_y = np_utils.to_categorical(Y_train)

下面的图片显示了Y_train和dummy_y实际上是什么。

我找不到任何可以帮助我的to_categorical文档。

提前致谢。

Answer 1

np.utils.to_categorical is used to convert array of labeled data(from 0 to nb_classes-1) to one-hot vector.

The official doc with an example.

In [1]: from keras.utils import np_utils # from keras import utils as np_utils
Using Theano backend.

In [2]: np_utils.to_categorical?
Signature: np_utils.to_categorical(y, num_classes=None)
Docstring:
Convert class vector (integers from 0 to nb_classes) to binary class matrix, for use with categorical_crossentropy.

# Arguments
    y: class vector to be converted into a matrix
    nb_classes: total number of classes

# Returns
    A binary matrix representation of the input.
File:      /usr/local/lib/python3.5/dist-packages/keras/utils/np_utils.py
Type:      function

In [3]: y_train = [1, 0, 3, 4, 5, 0, 2, 1]

In [4]: """ Assuming the labeled dataset has total six classes (0 to 5), y_train is the true label array """

In [5]: np_utils.to_categorical(y_train, num_classes=6)
Out[5]:
array([[ 0.,  1.,  0.,  0.,  0.,  0.],
       [ 1.,  0.,  0.,  0.,  0.,  0.],
       [ 0.,  0.,  0.,  1.,  0.,  0.],
       [ 0.,  0.,  0.,  0.,  1.,  0.],
       [ 0.,  0.,  0.,  0.,  0.,  1.],
       [ 1.,  0.,  0.,  0.,  0.,  0.],
       [ 0.,  0.,  1.,  0.,  0.,  0.],
       [ 0.,  1.,  0.,  0.,  0.,  0.]])

Answer 2

from keras.utils.np_utils import to_categorical

to_categorical(0, max_value_of_array)

假定类值是字符串，并且您将对其进行标签编码，因此每次从0到n个类开始。

for the same example:- consider an array of {1,2,3,4,2}

The output will be [zero value, one value, two value, three value, four value]

array([[ 0.,  1.,  0., 0., 0.],
       [ 0.,  0.,  1., 0., 0.],
       [ 0.,  0.,  0., 1., 0.],
       [ 0.,  0.,  0., 0., 1.],
       [ 0.,  0.,  1., 0., 0.]],

让我们看另一个示例：-

Again, for an array having 3 classes, Y = {4, 8, 9, 4, 9}

to_categorical(Y) will output

array([[0.,  0.,  0.,  0.,  1.,  0.,  0.,  0.,  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.,  1.,  0.,  0.,  0.,  0.,  0. ],
       [0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  1. ]]