Keras模型密集輸入形狀投擲誤差

Question

我有一個形狀為X_train.shape為(52, 54)的特征向量

當我訓練keras模型時，將錯誤拋出：

ValueError: Error when checking model input: expected dense_109_input to have shape (None, 52) but got array with shape (52, 54)

我已經嘗試了幾乎所有我能想到的一切以及掃描的堆棧溢出，但是我的問題仍然存在。 代碼如下：

import pandas as pd
from keras.models import Sequential
from keras.layers import Dense
from sklearn.model_selection import train_test_split
from sklearn.metrics import accuracy_score

##### Reading CSV #####  
data = pd.read_csv('Dataset/Emotion_data.csv')

X = data.ix[:, 4:]
y = data['label']

##### Normalizing #####
featureName = list(X)
for name in featureName:
    X[name] = (X[name] - min(X[name]))/(max(X[name]) - min(X[name]))

X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.3, random_state=3)

##### Model #####
model = Sequential()

model.add(Dense(100, input_shape=(54,), activation='relu'))
model.add(Dense(100, activation='relu'))
model.add(Dense(1, activation='softmax'))

model.compile(optimizer='adam', loss='mean_squared_error', metrics=['accuracy'])

model.fit(X_train, y_train)
prediction = model.predict(X_test)
print(accuracy_score(y_test, prediction))

如果有人對數據頭感興趣

In[42]: X_train.head()
Out[42]: 
       tempo  total_beats  average_beats  chroma_stft_mean  chroma_stft_std  \
35  0.438961     0.480897       0.505383          0.504320         0.938452   
34  0.520000     0.552580       0.500670          0.581778         0.680247   
63  0.477551     0.361328       0.334990          0.705472         0.357676   
27  0.477551     0.345419       0.309433          0.492245         0.728405   
43  0.520000     0.530305       0.495715          0.306097         0.663995   

    chroma_stft_var  chroma_cq_mean  chroma_cq_std  chroma_cq_var  \
35         0.932494        0.975206       0.394472       0.366960   
34         0.657810        0.654770       0.550766       0.522269   
63         0.333977        0.495473       0.618748       0.591578   
27         0.707998        0.644147       0.628125       0.601222   
43         0.640980        0.591299       0.639918       0.613379   

    chroma_cens_mean    ...       zcr_var  harm_mean  harm_std  harm_var  \
35          0.964034    ...      0.381363   0.021468  0.426776  0.225840   
34          0.755071    ...      0.213207   0.021598  0.115191  0.031476   
63          0.704930    ...      0.197960   0.021620  0.350194  0.163286   
27          0.715832    ...      0.247092   0.022253  0.319208  0.140714   
43          0.784991    ...      0.221276   0.021777  0.656981  0.471881   

    perc_mean  perc_std  perc_var  frame_mean  frame_std  frame_var  
35   0.362241  0.673257  0.467421    0.343459   0.174215   0.048846  
34   0.365434  0.152561  0.031588    0.091940   0.088991   0.018342  
63   0.340043  0.320664  0.116833    0.097610   0.077334   0.015154  
27   0.372315  0.604247  0.380492    0.995443   1.000000   1.000000  
43   0.377154  0.529161  0.296033    0.122519   0.089255   0.018417  

[5 rows x 54 columns]

Answer 1

您沒有在第一層中正確定義輸入形狀

model.add(Dense(100, input_shape=(54,), activation='relu'))

嘗試將第一層中的代碼更改為

model.add(Dense(100, input_shape=(52, 54), activation'relu))