about defining z in train_AAE.py

[minkyujeon]

Dear podgorskiy,

Thank you for your contribution to the open source community and academic. I was trying to use your code to perform novelty detection. Some codes of yours confused me, so I'd like to ask you If I'm wrong or misunderstanding something.

I'd like to ask about some lines in train_AAE.py. In line 173, variable z is defined as below.

z = torch.randn((batch_size, zsize)).view(-1, zsize, 1, 1)

Since x_fake comes from x, I think z should be defined as below

z = E(x)

I appreciate for your effort to review this code in advance.

[podgorskiy]

Please refer to eq. 9 from the paper.