nolearn.lasagne

Two introductory tutorials exist for nolearn.lasagne:

• Using convolutional neural nets to detect facial keypoints tutorial with code
• Training convolutional neural networks with nolearn

For specifics around classes and functions out of the lasagne package, such as layers, updates, and nonlinearities, you’ll want to look at the Lasagne project’s documentation.

nolearn.lasagne comes with a number of tests that demonstrate some of the more advanced features, such as networks with merge layers, and networks with multiple inputs.

Finally, there’s a few presentations and examples from around the web. Note that some of these might need a specific version of nolearn and Lasange to run:

• Oliver Dürr’s Convolutional Neural Nets II Hands On with code
• Roelof Pieters’ presentation Python for Image Understanding comes with nolearn.lasagne code examples
• Benjamin Bossan’s Otto Group Product Classification Challenge using nolearn/lasagne
• Kaggle’s instructions on how to set up an AWS GPU instance to run nolearn.lasagne and the facial keypoint detection tutorial
• An example convolutional autoencoder
• Winners of the saliency prediction task in the 2015 LSUN Challenge have published their lasagne/nolearn-based code.

API

class nolearn.lasagne.NeuralNet(layers, update=<function nesterov_momentum at 0x7f14b1fe9b70>, loss=None, objective=<function objective at 0x7f14b1f78510>, objective_loss_function=None, batch_iterator_train=<nolearn.lasagne.base.BatchIterator object at 0x7f14b1f745f8>, batch_iterator_test=<nolearn.lasagne.base.BatchIterator object at 0x7f14b1f74668>, regression=False, max_epochs=100, train_split=<nolearn.lasagne.base.TrainSplit object at 0x7f14b1f746a0>, custom_score=None, X_tensor_type=None, y_tensor_type=None, use_label_encoder=False, on_epoch_finished=None, on_training_started=None, on_training_finished=None, more_params=None, verbose=0, **kwargs)

A scikit-learn estimator based on Lasagne.

class nolearn.lasagne.BatchIterator(batch_size)

Standard batch iterator.

Given a reference to the X data and y target variables, it’s the batch iterator’s job to generate batches from these in (__iter__()).

Here’s a simple example:

>>> X, y = np.arange(18).reshape((9, 2)), np.arange(9)
>>> bi = BatchIterator(batch_size=3)
>>> for Xb, yb in bi(X, y):
...     print(Xb)
...     print(yb)
[[0 1]
 [2 3]
 [4 5]]
[0 1 2]
[[ 6  7]
 [ 8  9]
 [10 11]]
[3 4 5]
[[12 13]
 [14 15]
 [16 17]]
[6 7 8]

This implementation also knows how to deal with X inputs that are dictionaries mapping input names to arrays. Note that in this demo we’ll also only pass X, no y to the batch iterator, which also works:

>>> X1 = np.arange(18).reshape((9, 2))
>>> X2 = np.arange(18)[::-1].reshape((9, 2))
>>> bi = BatchIterator(batch_size=3)
>>> for Xb, yb in bi({'X1': X1, 'X2': X2}):
...     pass
>>> print(Xb['X1'])
[[12 13]
 [14 15]
 [16 17]]
>>> print(Xb['X2'])
[[5 4]
 [3 2]
 [1 0]]
>>> assert yb is None

class nolearn.lasagne.TrainSplit(eval_size, stratify=True)