Keras/Tensorflow中简单神经网络的误差

Question

我正在建立一个简单的神经网络。数据是一个231长的向量，它是一个热编码的向量。每个231个长向量被分配一个8个长的热编码标签。

到目前为止，我的代码是：

ssdf = pd.read_csv("/some/path/to/1AMX_one_hot.csv", sep=',')

ss = ssdf.iloc[:,3:11] # slice the df for the ss
labels = ss.values # vector of all ss's
labels = labels.astype('int32')
# data
onehot = ssdf.iloc[:,11:260]
data = onehot.values
data = data.astype('int32')

model = tf.keras.Sequential()
# Adds a densely-connected layer with 64 units to the model:
model.add(layers.Dense(64, activation='relu'))

# Add another:
model.add(layers.Dense(64, activation='relu'))

# Add a softmax layer with 8 output units:
model.add(layers.Dense(8, activation='softmax'))


model.compile(Adam(lr=.0001), 
          loss='sparse_categorical_crossentropy', 
          metrics=['accuracy']
)

## fit the model
model.fit(data, labels, epochs=10, batch_size=32)

问题是输出层是8个单位，但是我的标签不是单个单位，它们是一个热编码的8个长矢量。如何将此表示为输出？

错误消息是：

无法用非浮点dtype构建密集层

完全回溯：

Traceback (most recent call last):
  File "/some/path/to/file/main.py", line 36, in <module>
    model.fit(data, labels, epochs=10, batch_size=32)
  File "/anaconda3/lib/python3.7/site-    packages/tensorflow/python/keras/engine/training.py", line 806, in fit
    shuffle=shuffle)
  File "/anaconda3/lib/python3.7/site-    packages/tensorflow/python/keras/engine/training.py", line 2503, in     _standardize_user_data
    self._set_inputs(cast_inputs)
  File "/anaconda3/lib/python3.7/site-    packages/tensorflow/python/training/tracking/base.py", line 456, in     _method_wrapper
    result = method(self, *args, **kwargs)
  File "/anaconda3/lib/python3.7/site-    packages/tensorflow/python/keras/engine/training.py", line 2773, in     _set_inputs
    outputs = self.call(inputs, training=training)
  File "/anaconda3/lib/python3.7/site-    packages/tensorflow/python/keras/engine/sequential.py", line 256, in call
outputs = layer(inputs, **kwargs)
  File "/anaconda3/lib/python3.7/site-    packages/tensorflow/python/keras/engine/base_layer.py", line 594, in     __call__
    self._maybe_build(inputs)
  File "/anaconda3/lib/python3.7/site-    packages/tensorflow/python/keras/engine/base_layer.py", line 1713, in     _maybe_build
    self.build(input_shapes)
  File "/anaconda3/lib/python3.7/site-    packages/tensorflow/python/keras/layers/core.py", line 963, in build
    'dtype %s' % (dtype,))

Answer 1

示例代码中有几个问题：

1. 您的网络需要输入层或输入形状。
2. 将数据和标签输入为：astype（np.float32）

如果你的标签是形状（150,8），然后用8个神经元来拟合最后一层。

model.add(layers.Dense(8, activation='softmax'))
model.compile(Adam(lr=0.0001),
              loss='categorical_crossentropy',
              metrics=['accuracy'])

更新：

ssdf = pd.read_csv("/some/path/to/1AMX_one_hot.csv", sep=',')

ss = ssdf.iloc[:,3:11] # slice the df for the ss
labels = ss.values # vector of all ss's
labels = labels.astype('float32')                     # changed this
# data
onehot = ssdf.iloc[:,11:260]
data = onehot.values
data = data.astype('float32')                         # changed this

model = tf.keras.Sequential()
# Adds a densely-connected layer with 64 units to the model:
model.add(layers.Dense(64, activation='relu'))

# Add another:
model.add(layers.Dense(64, activation='relu'))

# Add a softmax layer with 8 output units:
model.add(layers.Dense(8, activation='softmax'))


model.compile(Adam(lr=.0001), 
          loss='categorical_crossentropy',            # changed this
          metrics=['accuracy']
)

## fit the model
model.fit(data, labels, epochs=10, batch_size=32)