Question

我创建了一个Conv1D模型用于文本分类。

在最后一次密集使用softmax / Sigmoid时，结果为

softmax => [0.98502016 0.0149798 ]
sigmoid => [0.03902826 0.00037046]

我只希望S型结果的第一个索引应该至少大于0.8 。只希望多类应该有独立的结果。我该如何实现？

模型摘要：

_________________________________________________________________
Layer (type)                 Output Shape              Param #   
=================================================================
embedding (Embedding)        (None, 128, 100)          600       
_________________________________________________________________
conv1d (Conv1D)              (None, 126, 128)          38528     
_________________________________________________________________
max_pooling1d (MaxPooling1D) (None, 63, 128)           0         
_________________________________________________________________
conv1d_1 (Conv1D)            (None, 61, 128)           49280     
_________________________________________________________________
max_pooling1d_1 (MaxPooling1 (None, 30, 128)           0         
_________________________________________________________________
conv1d_2 (Conv1D)            (None, 28, 128)           49280     
_________________________________________________________________
max_pooling1d_2 (MaxPooling1 (None, 14, 128)           0         
_________________________________________________________________
flatten (Flatten)            (None, 1792)              0         
_________________________________________________________________
dense (Dense)                (None, 2)                 3586      
=================================================================
Total params: 141,274
Trainable params: 141,274
Non-trainable params: 0
_________________________________________________________________
model.add(keras.layers.Dense(num_class, activation='sigmoid'))
model.compile(loss='categorical_crossentropy',
              optimizer='rmsprop', metrics=['acc'])

Answer 1

我同意@ blue-phoenox的评论，即您不应该将Sigmoid与交叉熵一起使用，因为类的概率之和不等于1。但是，如果您有使用ReferenceError: "email" is not defined. (line 7, file "Code")的理由，则可以通过矢量元素的总和对输出进行归一化，使其等于1：

sigmoid

您会得到：

output = output/tf.reshape(tf.reduce_sum(output, 1), (-1, 1))

要在import tensorflow as tf output = tf.Variable([[0.03902826, 0.00037046]]) output = output/tf.reshape(tf.reduce_sum(output, 1), (-1, 1)) summedup = tf.reduce_sum(output, axis=1) with tf.Session() as sess: sess.run(tf.global_variables_initializer()) print(output.eval()) # [[0.9905971 0.00940284]] - new output print(summedup.eval()) # [1.] - summs up to 1中实现它，您可以像这样创建keras的子类：

tf.keras.layers.Layer

然后在您的from tensorflow.keras import layers class NormLayer(layers.Layer): def __init__(self): super(NormLayer, self).__init__() def call(self, inputs): return inputs / tf.reshape(tf.reduce_sum(inputs, 1), (-1, 1))模型中使用它：

Sequential()

Answer 2

Sigmoid产生0到1之间的输出。如果对softmax和Sigmoid使用相同的损失函数，则它将不起作用。尝试使用binary_crossentropy。而且，如果您有两个以上的班级，我认为您不会在寻找S型。

增加S型预测输出值？

2 个答案: