2类以上的Tensorflow神经网络

Question

所以我在udemy上看了一个tensorflow的教程，决定自己试试，他说如果你想要2个以上的类别，就把激活改为 "softmax"，单位改为4，因为我有4个不同的类别（从0:1改为1:4），如果 "y "中只有2个不同的值，一切都能正常工作，但是一旦我把它改为4个单位和4个类别，我就会得到错误。

ValueError: 检查目标时出错：预期dense_3的形状为(4,)，但得到的是形状为(1,)的数组。

即使改回 "1 "的形状，结果也是真假不分。

我的数据集在Y中。

import numpy as np

dataset = np.load('/Users/alex/desktop/ANN_dataset_for_brightness.npy')
X = dataset[:, 0:17]
y = dataset[:, 17:19]

for i in range (27):
    if y[i] == 400:
        y[i] = 4
    elif y[i] == 300:
        y[i] = 3
    elif y[i] == 200:
        y[i] = 2
    elif y[i] == 100:
        y[i] = 1




from sklearn.model_selection import train_test_split
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size = 0.2, random_state = 0)


# Importing the Keras libraries and packages
from keras.models import Sequential
from keras.layers import Dense

# Initialising the ANN
classifier = Sequential()

# Adding the input layer and the first hidden layer // input dim for input layer
classifier.add(Dense(activation="relu", input_dim=17, units=6, kernel_initializer="uniform"))

# Adding the second hidden layer
classifier.add(Dense(activation="relu", units=6, kernel_initializer="uniform"))

这里有问题

# units = categories, softmax = more than 2
classifier.add(Dense(activation="softmax", units=4, kernel_initializer="uniform"))







# Compiling the ANN
classifier.compile(optimizer = 'adam', loss = 'binary_crossentropy', metrics = ['accuracy'])

# Fitting the ANN to the Training set
classifier.fit(X_train, y_train, batch_size = 27, nb_epoch = 100)

# Part 3 - Making the predictions and evaluating the model

# Predicting the Test set results
y_pred = classifier.predict(X_test)
y_pred = (y_pred > 0.5)

# Making the Confusion Matrix
from sklearn.metrics import confusion_matrix
cm = confusion_matrix(y_test, y_pred)

Answer 1

与多类任务，你的y_train和y_test必须是一个热编码。这意味着它们必须有维度(number_of_samples, 4)，其中4表示类的数量。

你需要在训练-测试拆分之前对它们应用 tensorflow.keras.utils.to_categorical。

y = to_categorical(y, 4)

参考文献 https:/www.tensorflow.orgapi_docspythontfkerasutilsto_categorical