How can I get hidden layer representation of the given data

[ohuole233]

When deep belief network is implemented for representation learning, I'm confused about the representation of hidden layers for the original data matrix.
The method sigmoid_layers[-1].output seems doesn't work with no representation for the matrix acquired except 0.
Has anybody encountered such confusion?

[tingjianlau]

In DBN.py, I add the extract_features() to get the output of the rbms.

def extract_features(self, aim_set_x, batch_size):
        n_aim_batches = aim_set_x.get_value(borrow=True).shape[0]
        n_aim_batches //= batch_size
        index = T.lscalar('index')  # index to a [mini]batch

         # extract the feature
        extract_features_i = theano.function(
            [index],
            self.sigmoid_layers[-1].output,
            givens={
                self.x: aim_set_x[
                    index * batch_size: (index + 1) * batch_size
                ]
            }
        )      
        return [extract_features_i(i) for i in range(n_aim_batches)]