Classification with 3D Gans

[adrianxsalazar]

Dear meetshah1995,

I do not understand a few things about the model and the implementation and I would like to ask you some questions.

1. In the discriminator, when you generate the features for training by taking the responses of the second, third and fourth cnn layers. Do you take just the responses of the real instances or the responses of the generated and real instances?

2)Do you use the mid-level features produced by the discriminator at the beginning of the training process or you just use the mid-level features once the model is trained?

3)After training the model and producing all the latest features for the 3D classifier. How do you transform new instances that we want to classify into the same format as the mid levels features extracted from the discriminator? This is, how can classify new 3D instances?

And these are my questions. I will appreciate a lot if you decide to spend part of your time to reply to my message. It will solve a lot of problems.

Kind regards,

Adrian Salazar Gomez