如何解决张量流Keras中的Conv2D模型
我尝试用大小为（2000，20，10，12）的X_train和大小为（2000，12）的y_train馈送Conv2D网络-每个通道一个输出，因为它是多对多分类。 我建立我的模型（下面的代码）并毫无问题地进行编译。 但是，当我想用X_train和y_train填充它时，虽然我给出了正确的输入大小，但是该模型给出了一条错误消息：
ValueError: ('Error when checking model target: expected no data,but got:
',array([[ 1.,-1.,...,1.,1.],[ 1.,1.]]))
这里是代码（简体）：
### Part 0. Import
import numpy as np
import pandas as pd
import tensorflow as tf
from tensorflow import keras
import matplotlib.pyplot as plt
from matplotlib import rc
from tensorflow.keras.layers import Dropout,Activation,Dense
from tensorflow.keras.layers import LSTM,Flatten,Add,TimeDistributed,Conv2D
from tensorflow.keras.models import Sequential
from tensorflow.keras import Model
from sklearn.metrics import confusion_matrix
from util_model import * # own code for data management
from util_LSTM import * # own code for organizing the X matrix into 3D matrix for LSTM model
### Part 1. Assumptions about the data
horiz = 1 ### time horizon of the investment in days ###
seq_length = 20 ### number of days for enriching the LSTM
step = 1 ### time lag within LSTM memory batch
eval_trigger = 0.001
nn_start = 250 ### initial value of X and Y matrices out of the total dataset
nn_size = 2500 ### length of the X & Y matrices starting from lstmStart index
proportionTrain = 0.8
### Part 2. Organizing the data into X_train,X_test,y_train & y_test for Conv2D
dataX,dataY = get_model_data(dbInput,dbList,ric,horiz)
Xs,ys = LSTM_create_dataset(dataX,dataY,seq_length,step) # slide over seq_length for a 3D matrix
Xslen = len(Xs)
if i == 0:
yconv.append(ys)
Xconv.append(Xs)
minlen = Xslen
i += 1
else:
i += 1
if minlen == Xslen:
yconv.append(ys)
Xconv.append(Xs)
else:
print('data ',i,' not added for size mismatch')
Xconv = np.array(Xconv)
yconv = np.array(yconv)
Xconv = np.reshape(Xconv,(Xconv.shape[1],Xconv.shape[2],Xconv.shape[3],Xconv.shape[0]))
yconv = np.reshape(yconv,(yconv.shape[1],yconv.shape[2],yconv.shape[0]))
## Xconv is a 4D matrix and yconv is a 2D matrix
(X_train,y_train),(X_test,y_test),(res_train,res_test) = CONV_train_test_return(Xconv,yconv,nn_start,nn_size,proportionTrain)
(X_train,X_test),(train_mean,train_std) = get_CONV_minmaxscaleX(X_train,X_test)
### Part 3. Model assumptions
validation_split = 0.1
model = keras.Sequential()
dropout = 0.2
optimizer = 'adam' ### Optimizer of the compiled model
learning = 0.001
loss = 'mean_squared_error'
verbose = 1 ## 0 = hidden computation // 1 = computation printed
batch_size = 128
epochs = 15
layer_1 = 128
layer_2 = 64
activ1 = 'relu'
activ2 = 'tanh'
activ3 = 'hard_sigmoid'
# available layers
layer_maxpool2D = tf.keras.layers.MaxPooling2D(pool_size=(2,2),padding = 'valid',strides= None,input_shape=(X_train.shape[1],X_train.shape[2],X_train.shape[3]))
layer_dense1 = Dense(units= layer_1,activation = 'relu')
layer_dense2 = Dense(units= layer_2,activation = activ1)
layer_LSTM1 = keras.layers.LSTM(units=layer_1,activation = 'relu',return_sequences = False,X_train.shape[2]))
layer_LSTMstack1 = keras.layers.LSTM(units=layer_2,return_sequences = True,X_train.shape[2]))
layer_LSTMstack2 = keras.layers.LSTM(units=layer_2,return_sequences = True)
layer_LSTMstackend = keras.layers.LSTM(units=layer_2,activation = 'relu')
layer_conv2D1 = keras.layers.Conv2D(filters = 12,kernel_size= (3,3),padding = 'same',X_train.shape[3]))
layer_output = Dense(units = 1)
# Model architecture
#model.add(TimeDistributed(Conv2D(filters = 12,X_train.shape[3]))))
model.add(Conv2D(filters = 12,X_train.shape[3])))
model.add(layer_maxpool2D)
model.add(Flatten())
model.add(layer_dense2)
model.add(layer_output)
### Part 4. Model compile & Train
model.compile(loss = loss,optimizer = optimizer)
history = model.fit(X_train,y_train,epochs = epochs,batch_size = batch_size,validation_split = validation_split,verbose = verbose)
模型到此停止，并显示上面共享的错误消息。 知道为什么以及如何解决吗？ 提前谢谢
版权声明：本文内容由互联网用户自发贡献，该文观点与技术仅代表作者本人。本站仅提供信息存储空间服务，不拥有所有权，不承担相关法律责任。如发现本站有涉嫌侵权/违法违规的内容， 请发送邮件至 dio@foxmail.com 举报，一经查实，本站将立刻删除。