No backend sub-module imported

[sxhfut]

Traceback (most recent call last):
File "/Users/sunxiao/PycharmProjects/testscikit/testscikit.py", line 116, in
clf.fit(train_X, train_y)
File "/Users/sunxiao/myproject/scikit-neuralnetwork/sknn/mlp.py", line 259, in fit
return super(Classifier, self)._fit(X, yp)
File "/Users/sunxiao/myproject/scikit-neuralnetwork/sknn/mlp.py", line 129, in _fit
X, y = self._initialize(X, y)
File "/Users/sunxiao/myproject/scikit-neuralnetwork/sknn/mlp.py", line 42, in _initialize
self._backend = MultiLayerPerceptronBackend(self)
File "/Users/sunxiao/myproject/scikit-neuralnetwork/sknn/backend/init.py", line 12, in init
raise NotImplementedError("No backend sub-module imported.")
NotImplementedError: No backend sub-module imported.

Process finished with exit code 1

[sxhfut]

fixed after pip --upgrade scikit-neuralnetwork

[alexjc]

This needs to be fixed in the master branch, you just downgraded. Here's the workaround:

from sknn.backend import pylearn2

Also needs some documentation, I will leave this open as a reference until it's fixed in 0.3.

[rmkemker]

I am getting this error when I try to build a single AutoEncoder. Lasagne built correctly with sknn during the pip install, but I am still getting this error. Any thoughts on how to fix it? I am using Ubuntu 14.04.

[alexjc]

There is no auto-encoder implemented in the latest release for Lasagne. #129

You need to go back to 0.4 for the PyLearn2 version, but that backend is not very well maintained and had many bugs.

[tgsmith61591]

So, is there any plan to tie the current AutoEncoder into any backend, or is it going to just be unusable for the forseeable future?