def main():
BATCH_SIZE = 100
EPOCH = 3
# load dataset
trainX, trainY, testX, testY, validX, validY = getDataXY()
model = build_model()
trainX, trainY = preDataStep(trainX, trainY)
validX, validY = preDataStep(validX, validY)
testX, testY = preDataStep(testX, testY)
# train model ...
print "train..."
earlyStopping = EarlyStopping(monitor='val_acc', patience=2)
print numpy.shape(testX)
model.fit(trainX,
trainY,
verbose=2,
shuffle=True,
callbacks=[earlyStopping],
nb_epoch=EPOCH,
validation_data=(validX, validY),
batch_size=BATCH_SIZE)
loss_and_metrics = model.evaluate(testX,
testY,
verbose=2,
batch_size=BATCH_SIZE)
print '==============='
print 'loss_metrics: ', loss_and_metrics
print numpy.shape(trainX)
return 0
import os
from time import time
from sklearn.externals import joblib
from sklearn.neural_network import MLPClassifier
from preprocessData import getDataXY
trainX, trainY, testX, testY, validX, validY = getDataXY()
print len(trainX), len(trainY), len(testX), len(testY), len(validX), len(
validY)
retrain = 1
X = trainX
y = trainY
# clf = SGDClassifier(loss="hinge", penalty="l2", n_iter=800)
# clf = tree.DecisionTreeClassifier()
# clf = NearestCentroid(metric='manhattan')
clf = MLPClassifier(solver='lbfgs',
alpha=1e-5,
hidden_layer_sizes=(13, 5),
random_state=1,
verbose=True,
max_iter=100,
early_stopping=True)
if not os.path.isfile('NNP.pkl') or retrain == 1:
clf.fit(X, y)
else:
clf = joblib.load('NNP.pkl')
