Hello author, I have a few questions about the loss function in your paper.
For NLL loss, it is required the z after inn to be standard normal, but in your code, it seems to be setting the z before feeding inn to be standard normal
Also I don't understand the meaning of the last term of the loss, the first two terms mean that the distribution of z before feeding into inn is consistent with the distribution of xa and xb
For datasets, how do I select attributes that share common features, e.g. I filtered the dataset of faces with glasses, can I use the same dataset of faces with glasses for xa and xb？
Hello author, I have a few questions about the loss function in your paper.
For NLL loss, it is required the z after inn to be standard normal, but in your code, it seems to be setting the z before feeding inn to be standard normal
Also I don't understand the meaning of the last term of the loss, the first two terms mean that the distribution of z before feeding into inn is consistent with the distribution of xa and xb
For datasets, how do I select attributes that share common features, e.g. I filtered the dataset of faces with glasses, can I use the same dataset of faces with glasses for xa and xb？