Question

我正在尝试将一维CNN放入我的数据中。数据由MEL频率组成，并具有以下特征：

X_train.shape = (68251, 99)
y_train_hot.shape = (68251, 35)<-- one hot encoding with 35 output classes

当我尝试训练模型时，以下代码出现此值错误：

ValueError: Error when checking input: expected conv1d_5_input to have 3 dimensions, but got array with shape (68251, 99)


#hyperparameters
input_dimension = 68251
learning_rate = 0.0025
momentum = 0.85
hidden_initializer = random_uniform(seed=1)
dropout_rate = 0.2
# create model
model = Sequential()
model.add(Convolution1D(nb_filter=32, filter_length=3, input_shape=X_train.shape, activation='relu'))
model.add(Convolution1D(nb_filter=16, filter_length=1, activation='relu'))
model.add(Flatten())
model.add(Dropout(dropout_rate))
model.add(Dense(128, input_dim=input_dimension, kernel_initializer=hidden_initializer, activation='relu'))
model.add(Dropout(dropout_rate))
model.add(Dense(64, kernel_initializer=hidden_initializer, activation='relu'))
model.add(Dense(2, kernel_initializer=hidden_initializer, activation='softmax'))

sgd = SGD(lr=learning_rate, momentum=momentum)
model.compile(loss='binary_crossentropy', optimizer=sgd, metrics=['acc'])
model.fit(X_train, y_train_hot, epochs=5, batch_size=128)
predictions = model.predict_proba(X_test)

ans = pd.DataFrame(predictions)
ans = ans[0]

当我将X_train和X_test分别重塑为（68251，99，1）和（17063，99，1）时，出现以下错误：

ValueError: Input 0 is incompatible with layer conv1d_7: expected ndim=3, found ndim=4

Answer 1

编辑
这次我实际上编译了您的模型，发现了一些问题并纠正了这些问题：

# create model
model = Sequential()
model.add(Convolution1D(nb_filter=32, filter_length=3, input_shape=(99, 1), activation='relu'))
model.add(Convolution1D(nb_filter=16, filter_length=1, activation='relu'))
model.add(Flatten())
model.add(Dropout(0.2))
model.add(Dense(128, activation='relu'))
model.add(Dropout(0.2))
model.add(Dense(64, activation='relu'))
model.add(Dense(35, activation='softmax'))


model.compile(loss='binary_crossentropy', optimizer='sgd', metrics=['acc'])
model.fit(X_train, y_train_hot, epochs=5, batch_size=128)

您的输出为2，应为35，因为您有35个输出类，您的第一个Dense不需要input_dim，因为形状将由上一层推断出来，并且输入形状不正确。
希望这会有所帮助

Answer 2

尝试更改

model.add(Convolution1D(nb_filter=32, filter_length=3, input_shape=X_train.shape, activation='relu'))

到

model.add(Convolution1D(nb_filter=32, filter_length=3, input_shape=(99,), activation='relu'))

并将X_train传递给fit函数而无需重塑

1D CNN keras错误的形状

2 个答案: