def training(d):
"""
Builds a network and trains it.
"""
print("Creating Neural Network")
n = NNregression(d)
print("Setting Up Neural Network")
n.setupNN()
print("Training Neural Network")
n.runTraining()
print("Returning Trainier")
return n.Trainer
class NNMatcher(LinearMatcher):
def __init__(self):
LinearMatcher.__init__(self)
def train(self, pairings):
self.dataset = SupervisedDataSet(self.cheese_feat_len, self.wine_feat_len)
for cheese, wine in pairings:
cheese_desc = self.cheeses[cheese]
wine_desc = self.wines[wine]
self.dataset.addSample(cheese_desc, wine_desc)
self.nn = NNregression(self.dataset, hidden=self.cheese_feat_len + self.wine_feat_len, maxepochs=35)
self.nn.setupNN()
self.nn.runTraining()
def predict_feat(self, cheese_desc):
return self.nn.Trainer.module.activate(cheese_desc)
def train(self, pairings):
self.dataset = SupervisedDataSet(self.cheese_feat_len, self.wine_feat_len)
for cheese, wine in pairings:
cheese_desc = self.cheeses[cheese]
wine_desc = self.wines[wine]
self.dataset.addSample(cheese_desc, wine_desc)
self.nn = NNregression(self.dataset, hidden=self.cheese_feat_len + self.wine_feat_len, maxepochs=35)
self.nn.setupNN()
self.nn.runTraining()
def train_network(train,target):
"""
Trains via linear regression from the target and train data
Arguments:
train : an array of training data
target: an array of associated target data
Returns:
The trained model
"""
print("Setting up Neural Network data with %s train features and %s targets sets" % (len(train[0]), len(target[0])))
data = SupervisedDataSet(len(train[0]),len(target[0]))
data.setField('input', train)
data.setField('target', target)
n = NNregression(data)
n.setupNN()
print("Training Neural Network on %s training sets" % len(data))
n.runTraining()
return n.Trainer.module
def run(self, ds_train, ds_test):
"""
This function evaluates the ANN using a test set and has the error-rate as an output.
Args:
:param ds_train (TweetClassificationDatasetFactory): the training dataset the neural network is trained with.
:param ds_test (TweetClassificationDatasetFactory): the test dataset evaluated.
:returns error (float): the percent error of the test dataset, tested on the neural network.
"""
self.network = NNregression(ds_train)
self.network.setupNN(hidden=self.hidden, verbose=True)
self.network.Trainer.trainUntilConvergence(
dataset=ds_train,
maxEpochs=self.max_epochs,
continueEpochs=self.con_epochs)
error = self.test(ds_test)
return error
def run_with_crossvalidation(self, ds, iterations=5):
"""
This function estimates the performance of the neural network using crossvalidation using a specified dataset.
Args:
:param ds (TweetRegressionDatasetFactory): the dataset used to crossvalidate the network.
:param iterations (int, optional): number of iterations for the crossvalidation.
:returns error (float): the average percent error of the dataset, tested on the network using crossvalidation.
"""
x = ds['input']
y = ds['target']
n, m = x.shape
errors = np.zeros(iterations)
cv = cross_validation.KFold(n, iterations, shuffle=True)
i = 0
for train_index, test_index in cv:
x_train = x[train_index, :]
y_train = y[train_index, :]
x_test = x[test_index, :]
y_test = y[test_index, :]
ds_train = TweetClassificationDatasetFactory.convert_to_ds(x_train, y_train)
ds_test = TweetClassificationDatasetFactory.convert_to_ds(x_test, y_test)
self.network = NNregression(ds_train)
self.network.setupNN(hidden=self.hidden)
self.network.Trainer.trainUntilConvergence(
dataset=ds_train,
maxEpochs=self.max_epochs,
continueEpochs=self.con_epochs)
tstresult = self.test(ds_test)
errors[i] = tstresult[0]
i += 1
print "Simple Regression Neural Network cross-validation test errors: " % errors
return np.average(errors)
class SimpleRegressionNeuralNetwork(AI):
def __init__(self, hid_cnt=10, max_epochs=50, con_epochs=4):
self.hidden = hid_cnt
self.network = None
self.con_epochs = con_epochs
self.max_epochs = max_epochs
def run(self, ds_train, ds_test):
"""
This function evaluates the ANN using a test set and has the error-rate as an output.
Args:
:param ds_train (TweetClassificationDatasetFactory): the training dataset the neural network is trained with.
:param ds_test (TweetClassificationDatasetFactory): the test dataset evaluated.
:returns error (float): the percent error of the test dataset, tested on the neural network.
"""
self.network = NNregression(ds_train)
self.network.setupNN(hidden=self.hidden, verbose=True)
self.network.Trainer.trainUntilConvergence(
dataset=ds_train,
maxEpochs=self.max_epochs,
continueEpochs=self.con_epochs)
error = self.test(ds_test)
return error
def test(self, ds_test):
"""
This function evaluates the ANN using a test set and has the error-rate as an output.
Args:
:param ds_test (TweetRegressionDatasetFactory): the test dataset evaluated.
:returns error (float): the percent error of the test dataset, tested on the network.
"""
result = self.network.Trainer.module.activateOnDataset(ds_test)
error = mean_squared_error(ds_test['target'], result)
return error
def run_with_crossvalidation(self, ds, iterations=5):
"""
This function estimates the performance of the neural network using crossvalidation using a specified dataset.
Args:
:param ds (TweetRegressionDatasetFactory): the dataset used to crossvalidate the network.
:param iterations (int, optional): number of iterations for the crossvalidation.
:returns error (float): the average percent error of the dataset, tested on the network using crossvalidation.
"""
x = ds['input']
y = ds['target']
n, m = x.shape
errors = np.zeros(iterations)
cv = cross_validation.KFold(n, iterations, shuffle=True)
i = 0
for train_index, test_index in cv:
x_train = x[train_index, :]
y_train = y[train_index, :]
x_test = x[test_index, :]
y_test = y[test_index, :]
ds_train = TweetClassificationDatasetFactory.convert_to_ds(x_train, y_train)
ds_test = TweetClassificationDatasetFactory.convert_to_ds(x_test, y_test)
self.network = NNregression(ds_train)
self.network.setupNN(hidden=self.hidden)
self.network.Trainer.trainUntilConvergence(
dataset=ds_train,
maxEpochs=self.max_epochs,
continueEpochs=self.con_epochs)
tstresult = self.test(ds_test)
errors[i] = tstresult[0]
i += 1
print "Simple Regression Neural Network cross-validation test errors: " % errors
return np.average(errors)
def __call__(self, ds_train, ds_test):
return self.run(ds_train, ds_test)
def save(self, path):
"""
This function saves the neural network.
Args:
:param path (String): the path where the network is going to be saved.
"""
file_object = open(path, 'w')
pickle.dump(self.network, file_object)
file_object.close()
def load(self, path):
"""
This function loads the neural network.
Args:
:param path (String): the path where the neural network is going to be loaded from.
"""
file_object = open(path, 'r')
self.network = pickle.load(file_object)
def get_type(self):
"""
This function returns the type of problem and type of neural network.
:returns:
ProblemTypeEnum.Regression (enum): the type of problem.
AIEnum.SimpleRegressionNeuralNetwork (enum): the type of artificial intelligence.
"""
return ProblemTypeEnum.Regression, AIEnum.SimpleRegressionNeuralNetwork
def fill_with_predicted_data(self, ds, data):
"""
This function fills a dataset with the real and predicted values using a specified database.
Args:
:param ds (TweetRegressionDatasetFactory): the dataset used to
fill the dictionary with the real and predicted values.
:param data (dictionary): dataset gets filled with the real and the predicted values.
"""
target = ds['target']
out = self.network.Trainer.module.activateOnDataset(ds)
results = np.ravel(np.argmax(out, 1))
self.fill_data_regression(target, results, data)
''' Now lets build a neural network and train it using a back propagation network on the training data (trndata).
'''
fnn = buildNetwork(trndata.indim, 5, trndata.outdim, recurrent=False)
trainer = BackpropTrainer(fnn,
dataset=trndata,
momentum=0.1,
verbose=True,
weightdecay=0.01)
''' Originally I figured the best way to train was to do it myself, having looked at my code, and that in the doc, I think that it’s best to use the built-in functions:
'''
from pybrain.tools.neuralnets import NNregression, Trainer
# Create you dataset - as above
nn = NNregression(alldata)
nn.setupNN()
nn.runTraining()
''' Train the network for n epochs '''
# I am not sure about this, I don't think my production code is implemented like this
modval = ModuleValidator()
for i in range(1000):
trainer.trainEpochs(1)
trainer.trainOnDataset(dataset=trndata)
cv = CrossValidator(trainer, trndata, n_folds=5, valfunc=modval.MSE)
print "MSE %f @ %i" % (cv.validate(), i)
''' Now predict the test data again – this should really be a HOT set.
'''
print trndata['input'][0], trndata['target'][0], trndata['class'][0]
''' Now lets build a neural network and train it using a back propagation network on the training data (trndata).
'''
fnn = buildNetwork( trndata.indim, 5, trndata.outdim, recurrent=False )
trainer = BackpropTrainer( fnn, dataset=trndata, momentum=0.1, verbose=True, weightdecay=0.01 )
''' Originally I figured the best way to train was to do it myself, having looked at my code, and that in the doc, I think that it’s best to use the built-in functions:
'''
from pybrain.tools.neuralnets import NNregression, Trainer
# Create you dataset - as above
nn = NNregression(alldata)
nn.setupNN()
nn.runTraining()
''' Train the network for n epochs '''
# I am not sure about this, I don't think my production code is implemented like this
modval = ModuleValidator()
for i in range(1000):
trainer.trainEpochs(1)
trainer.trainOnDataset(dataset=trndata)
cv = CrossValidator( trainer, trndata, n_folds=5, valfunc=modval.MSE )
print "MSE %f @ %i" %( cv.validate(), i )
''' Now predict the test data again – this should really be a HOT set.