Can we perform Transfer learning? and Covolutional Autoencoder?

[always...@gmail.com]

Dear Sir

Sorry for bothering you with lots of questions.

Now i am using your code in my research.

My new questions are 

1) Can we do the transfer learning in your code?

2) Can we do the convolutional autoencoder in your code?

Thank you in advance

Best regard

Hang-Rai Kim

[Hagay Garty]

Hi, yes you can do both.

Autoencoders can be done using a specific network structure (specified in the config file) when one of the middle hidden layers has few outputs of size 1x1

By transfer learning, I guess you mean using pre-trained net weights. The network weights are saved in the 'net' variable returned from 'Train' 

You can re-use this variable to train on new data.

Hope it helps,

Hagay