def test_trainer_feedforward(self):
"""
Test: Feedforward
"""
t = trainer.Trainer()
t.train([101, 103], [201, 202, 203], 201)
print t.feedforward()
assert t.feedforward() == [
0.33506294978429246, 0.05512695704100274, 0.05512695704100274
]
def test_trainer_train(self):
"""
Test: Train
"""
t = trainer.Trainer()
t.train([101, 103], [201, 202, 203], 201)
rowin = [(101, 1, 0.516117), (103, 1, 0.516117)]
rowout = [(1, 201, 0.449819), (1, 202, 0.071222), (1, 203, 0.071222)]
for i, row in enumerate(
t.utils.db.con.execute('select * from inputs')):
assert row == rowin[i]
for i, row in enumerate(
t.utils.db.con.execute('select * from outputs')):
assert row == rowout[i]
def test_trainer_backpropagate(self):
"""
Test: Backpropagate
"""
t = trainer.Trainer()
t.train([101, 103], [201, 202, 203], 201)
targets = [0.0] * len([201, 202, 203])
targets[[201, 202, 203].index(201)] = 1.0
rowin = [(101, 1, 0.516117), (103, 1, 0.516117)]
rowout = [(1, 201, 0.449819), (1, 202, 0.071222), (1, 203, 0.071222)]
for i, row in enumerate(
t.utils.db.con.execute('select * from inputs')):
assert row == rowin[i]
for i, row in enumerate(
t.utils.db.con.execute('select * from outputs')):
assert row == rowout[i]
def test_trainer_object(self):
"""
Test: Trainer object
"""
t = trainer.Trainer()
assert t != None
def test_trainer_constructor(self):
"""
Test: Trainer constructor
"""
t = trainer.Trainer()
assert t.utils != None