Deep Learning Keras Framework notes of Autoencoder class

Deep learning Keras Frame Notes Autoencoder class use notes

　　

This is a very common auto-coding model for building. If the parameter is Output_reconstruction=true, then Dim (input) =dim (output), otherwise dim (output) =dim (hidden).

Inputshape: Depends on the definition of encoder

Outputshape: Depends on the definition of decoder

Parameters:

• Encoder: Encoder, which is a layer type or layer container type.
• Decoder: Decoder, which is a layer type or layer container type.
• Output_reconstruction:boolean. When the value is false, when the Predict () function is called, the output is the activation function that passes through the deepest hidden layer. Otherwise, the output of Thefinal decoder layer is presented. Be sure your validation the data conforms to thislogic if you decide to the use of any. (This piece is not quite understood, later to understand the additional)
• Weights: A list of numpy arrays that is used to initialize weights. This list has at least 1 elements, and its shape is (Input_dim, Output_dim).

Example:

From keras.layers import Containers     # Input shape: (nb_samples, +)  encoder =containers. Sequential ([Dense (input_dim=32), dense (8)])  decoder =containers. Sequential ([Dense (+, input_dim=8), dense (+)]) Autoencoder =sequential (     )  Autoencoder.add (Autoencoder ( Encoder=encoder, Decoder=decoder,output_reconstruction=false))  

By the way, a small advertisement, welcome to visit your own website: cylindrical template

Deep Learning Keras Framework notes of Autoencoder class