这个神经网络有多少个隐藏层和总层？

Question

我使用以下代码使用keras构建MLP。

model_relu = Sequential()
model_relu.add(Dense(256, activation='relu', input_shape=(input_dim,), kernel_initializer=RandomNormal(mean=0.0, stddev=0.062, seed=None)))
model_relu.add(Dense(128, activation='relu', kernel_initializer = RandomNormal(mean=0.0, stddev=0.125, seed=None)) )
model_relu.add(Dense(64, activation='relu', kernel_initializer = RandomNormal(mean=0.0, stddev=0.07, seed=None)) ) 
model_relu.add(Dense(output_dim, activation='softmax'))

model_relu.summary()

摘要是

Model: "sequential_19"
_________________________________________________________________
Layer (type)                 Output Shape              Param #   
=================================================================
dense_49 (Dense)             (None, 256)               200960    
_________________________________________________________________
dense_50 (Dense)             (None, 128)               32896     
_________________________________________________________________
dense_51 (Dense)             (None, 64)                8256      
_________________________________________________________________
dense_52 (Dense)             (None, 10)                650       

我想要这个MLP有多少个隐藏层。我们应该将此3称为隐藏层数还是4个隐藏层。 总层数是5（输入+ 3隐藏+ 1输出（softmax）吗？

Answer 1

您有1个input layer带有256个神经元，有2个hidden layers带有128和64个神经元，最后您有1个output layer带有10个神经元。