Can anyone explain how to do concatenation of flatten outputs from CNN with the outputs from DNN?

2 次查看（过去 30 天）
Wabi Demeke
Wabi Demeke 2020-7-23
评论： Poorya Khanali 2021-2-18
I have defined two layers
1 CNN_Layer which takes input size of 12*25*1 and implment successive Con2D filters, MaxPooling ,etc and finally flatten the inputs from the final layer.
2 DNN_Layer which takes input size of 4*1 and ouput the same as the input(Just takes raw input and give us the raw input without changing it) .
My goal is to concatenat the faltten outputs from CNN_Layers and the outputs from DNN_Layers for further training using FCC layers.
The output from CNN_Layer after flatten is dim 864, so I expact 868 after concatenation .
Is there any way we can do that in matlab?
I know how to do that in Python.
combined_model= concatenate([CNN_Layers.output, DNN_Layers.output])

请先登录，再回答此问题。

回答（1 个）

Mohammad Sami
Mohammad Sami 2020-7-23
Try using the concatenation Layer, introduced since R2019a.
https://www.mathworks.com/help/releases/R2020a/deeplearning/ref/nnet.cnn.layer.concatenationlayer.html
  1 个评论
Wabi Demeke
Wabi Demeke 2020-8-1
This is multiple inputs(input1 dim(12*25*1) & Input 2 dim(1*4*1)) problem. The data need to be combined before it feed into and I don't want that at the beginning( https://www.mathworks.com/help/deeplearning/ug/multiple-input-and-multiple-output-networks.html#mw_4f57107d-31b4-4432-8d4f-2065ce9008f0). I beleive concatenationlayer/depthconcatenationlayer works as long as the input layer takes input1 or input2.
Probably for the task I am looking is explaied here very well(https://www.mathworks.com/matlabcentral/fileexchange/74760-image-classification-using-cnn-with-multi-input-cnn)

请先登录，再进行评论。

类别

在 Help Center 和 File Exchange 中查找有关 Get Started with Statistics and Machine Learning Toolbox 的更多信息

标签

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by