multilabel classification using sigmoid layer

Hi,

I have multilabel problem with mutually non exclusive class labels ( chest x-rays) . I want to train a neural network for this . I have replaced softmax layer with sigmoid layer. now i get error: Missing softmax layer: A classification layer must be preseded by a softmax layer. Can anyone comment on this to help resolve this issue.

Thank you

[SC P]

I am also facing same problem. Can anyone comment on this to help resolve this issue?