我使用kera和cntk作为卷积神经网络的后端 . 列车数据的形状是(12459,49) . input_shape存在一些问题 . 输入是numpy数组 .
model= Sequential()
model.add(Conv1D(nb_filter=32,filter_length=3, strides=1,input_shape=
(None,x_train.shape),padding='same', dilation_rate=1, activation='relu',
use_bias=True, kernel_initializer='glorot_uniform',
bias_initializer='zeros',kernel_regularizer=None, bias_regularizer=None,
activity_regularizer=None, kernel_constraint=None, bias_constraint=None))
model.add(Conv1D(nb_filter=16, filter_length=1, strides=1,padding='same',
dilation_rate=1, activation='relu', use_bias=True,
kernel_initializer='glorot_uniform', bias_initializer='zeros',
kernel_regularizer=None, bias_regularizer=None, activity_regularizer=None,
kernel_constraint=None, bias_constraint=None))
model.add(MaxPooling1D(pool_size=2, strides=None, padding='valid'))
model.add(Dropout(dropout_rate))
model.add(Flatten())
错误:
input_variable(shape, dtype, needs_gradient, is_sparse, dynamic_axes, name)
3504 # TODO sparse for numpy arrays
3505
-> 3506 return input_variable(shape, is_sparse, dtype, needs_gradient,
name, dynamic_axes)
3507
3508 @typemap
OverflowError: in method 'input_variable', argument 3 of type '::CNTK::DataType'
它显示错误:在方法'input_variable'中,在第一个convlayer处类型为':: CNTK :: DataType'的参数3 . 我阅读了conv1D层的cntk input_variable和keras文档的文档 . 我无法在keras层中指定dtype,但根据我的理解,cntk在input_variable函数中需要dtype . 什么可以解决方案?