1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
|
#!/usr/bin/env python
"""General test for the NeuralNetwork libraries.
This exercises various elements of the BackPropagation NeuralNetwork
libraries.
"""
# standard library
import random
import unittest
# local stuff
from Bio.NeuralNetwork.Training import TrainingExample, ExampleManager
from Bio.NeuralNetwork.StopTraining import ValidationIncreaseStop
class StopTrainingTest(unittest.TestCase):
"""Test functionality for stopping training networks.
"""
def test_validation_increase_stop(self):
"""Stop training when the ValidationExamples increase.
"""
stopper = ValidationIncreaseStop(max_iterations = 20,
min_iterations = 2)
stopper.last_error = 1.0
do_stop = stopper.stopping_criteria(5, 1.0, 1.5)
assert do_stop == 1, \
"Did not tell us to stop when validation error increased."
stopper.last_error = 1.0
do_stop = stopper.stopping_criteria(1, 1.0, 1.5)
assert do_stop == 0, \
"Told us to stop before we reached the minimum iterations."
stopper.last_error = 1.0
do_stop = stopper.stopping_criteria(25, 1.0, 0.5)
assert do_stop == 1, \
"Did not tell us to stop when reaching maximum iterations."
class ExampleManagerTest(unittest.TestCase):
"""Tests to make sure the example manager is working properly.
"""
def setUp(self):
self.num_examples = 500
self.examples = []
for make_example in range(self.num_examples):
inputs = []
for input_make in range(3):
inputs.append(random.randrange(1, 7))
outputs = [random.randrange(1, 7)]
self.examples.append(TrainingExample(inputs, outputs))
def test_adding_examples(self):
"""Make sure test examples are added properly.
"""
manager = ExampleManager()
# figure out the expected number of examples in each category
expected_train = manager.training_percent * self.num_examples
expected_validation = manager.validation_percent * self.num_examples
expected_test = self.num_examples - expected_train \
- expected_validation
manager.add_examples(self.examples)
for expect, actual in [(expected_train, len(manager.train_examples)),
(expected_validation,
len(manager.validation_examples)),
(expected_test, len(manager.test_examples))]:
wrong_percent = abs(expect - actual) / self.num_examples
assert wrong_percent < .1, \
"Deviation in how examples were added, expect %s, got %s" \
% (expect, actual)
def test_partioning_examples(self):
"""Test that we can change how to partition the test examples.
"""
manager = ExampleManager(0, 0)
manager.add_examples(self.examples)
assert len(manager.test_examples) == self.num_examples, \
"Did not partion correctly to test_examples."
manager = ExampleManager(1.0, 0)
manager.add_examples(self.examples)
assert len(manager.train_examples) == self.num_examples, \
"Did not partition correctly to train_examples."
manager = ExampleManager(0, 1.0)
manager.add_examples(self.examples)
assert len(manager.validation_examples) == self.num_examples, \
"Did not partition correctly to validation_examples."
if __name__ == "__main__":
runner = unittest.TextTestRunner(verbosity = 2)
unittest.main(testRunner=runner)
|
