Implement MLP

[dwettstein]

Use the provided training set to train an MLP with one hidden layer. Apply the trained MLP to classify the test set. Perform cross-validation with the following parameters:

• Optimize number of neurons in the hidden layer (typically in the range [10,100]).
• Optimize learning rates (typically in the range [0.1,1])
• Optimize number of training iterations. Plot a graph showing the error on the training set and the validation set, respectively, with respect to the training epochs.
• Perform the random initialization several times and choose the best network during cross- validation.

https://pypi.python.org/pypi/bob.learn.mlp

[dwettstein]

Use MLP implementation of scikit-learn: https://github.com/scikit-learn/scikit-learn/blob/master/sklearn/neural_network/multilayer_perceptron.py

[dwettstein]

Alternative: https://pypi.python.org/pypi/scikit-neuralnetwork

[dwettstein]

Done with 2f010ff173e2abc25fe4d28ff4e21b1c645e474e