svhn semi-supervised learning vector dimensionality

[yatindandi]

The original ALI paper mentions 'The last three hidden layers of the encoder as well as its output are concatenated to form a 8960-dimensional feature vector.' Concatenating the output and hidden layers as you did in 'test_semisup.py' leads to a vector with lower dimensionality. Is this simply a typo in the original paper?

[9310gaurav]

I had this concern when I implemented the paper. The architecture is the same as mentioned in their paper and in their released Theano code. Better to check with authors but I think it's a typo in the paper.