def get_density_with_ann(self):
obj = libfann.fann_create_standard_array(2, (3, 1))
ann = fann.fann_class(obj)
patterns = fann.read_train_from_file("train_data.txt")
ann.train_on_data(patterns, 1000, 1, 0.0)
return [
line[3]
for line in map(lambda x: [x.density, x.bytes, len(x.text), x.text], self.lines)
if ann.run(line[0:2])[0] > 0
]
def fann_test(trainingText):
""" The function to using fast artifitial neural network."""
# using "fann" package in machine learning
from pyfann import fann, libfann
# This creates a new single-layer perceptron with 1 output and 3 inputs.
obj = libfann.fann_create_standard_array(2, (3, 1))
ann = fann.fann_class(obj)
# Load the data we described above.
patterns = fann.read_train_from_file('training.txt')
ann.train_on_data(patterns, 1000, 1, 0.0)
# Then test it with different data.
for datin, datout in validation_data:
result = ann.run(datin)
print 'Got:', result, ' Expected:', datout
def main():
if (len(sys.argv) != 3):
print "bad number of argument. got: " + str(len(sys.argv)) + " expected: 3"
print "usage: extract_text.py <newswatch-db-file> <training_file>"
sys.exit(1)
page = load_page(load_html(sys.argv[1]))
manual_training(page, sys.argv[2])
# This creates a new single-layer perceptron with 1 output and 3 inputs.
obj = libfann.fann_create_standard_array(2, (3, 1))
ann = fann.fann_class(obj)
# Load the data we described above.
if (os.path.exists(sys.argv[2]) == True):
patterns = fann.read_train_from_file(sys.argv[2])
ann.train_on_data(patterns, 1000, 1, 0.0)
# Then test it with different data.
for index, line in enumerate(page.lines):
if (index == 0):
input = [0.0, 0.0, 0.0]
else:
input = [page.lines[index - 1].density,
page.lines[index - 1].bytes,
len(page.lines[index - 1].text)]
input = [line.density. line.bytes, len(line.text)]
if (index + 1 >= len(page.lines) - 1):
input = [0.0, 0.0, 0.0]
else:
input = [page.lines[index + 1].density,
page.lines[index + 1].bytes,
len(page.lines[index + 1].text)]
result = ann.run(input)
print line.text + "\n" + 'Got:', result
