在顺序Keras模型中将一维数据加载到密集层中

Question

I have the results of a trained model, ending in a Flatten layer in numpy output files. I try to load them and use them as inputs of a Dense layer.

train_data = np.load(open('bottleneck_flat_features_train.npy'))
train_labels = np.array([0] * (nb_train_samples / 2) + [1] * (nb_train_samples / 2))
#
validation_data = np.load(open('bottleneck_flat_features_validation.npy'))
validation_labels = np.array([0] * (nb_validation_samples / 2) + [1] * (nb_validation_samples / 2))
#
top_m  = Sequential()
top_m.add(Dense(2,input_shape=train_data.shape[1:], activation='sigmoid', name='top_dense1'))
top_m.compile(optimizer='rmsprop', loss='categorical_crossentropy', metrics=['accuracy'])
#
top_m.fit(train_data, train_labels,
    nb_epoch=nb_epoch, batch_size=my_batch_size,
    validation_data=(validation_data, validation_labels))

However I get the following error message:

    ValueError: Error when checking model target: expected top_dense1 to have
 shape (None, 2) but got array with shape (13, 1)

My input dimensions are (16,1536) - 16 images for this limited trail run, 1536 features.

>>> train_data.shape
(16, 1536)

The dense layer should expect a one dimensional 1536 long array.

>>> train_data.shape[1]
1536

What should I do? Many thanks!

Answer 1

您是否不想编译并拟合top_m而不是模型？

Answer 2

I have found my problem - I did not define the lables correctly. I have switched the model compilation to a sparse categorical crossentropy mode.

my current code is

def train_top_model():
    train_data = np.load(open('bottleneck_flat_features_train.npy'))
    train_labels = np.array([0] * (nb_train_samples / 2) + [1] * (nb_train_samples / 2))
#
    validation_data = np.load(open('bottleneck_flat_features_validation.npy'))
    validation_labels = np.array([0] * (nb_validation_samples / 2) + [1] * (nb_validation_samples / 2))
#
    top_m  = Sequential()
    top_m.add(Dense(2,input_shape=train_data.shape[1:], activation='softmax', name='top_dense1'))
    top_m.compile(optimizer='rmsprop', loss='sparse_categorical_crossentropy', metrics=['accuracy'])
#
    top_m.fit(train_data, train_labels,
    nb_epoch=nb_epoch, batch_size=my_batch_size,
    validation_data=(validation_data, validation_labels))

Now it works and converges.