def test_train_with_delays(self):
n = Neuron(S0 = 16, H = 4.0, G = 2.0, C = 1, D1 = 2, D2 = 2)
# Fix neuron delays manually for the test
n.synapses['delay'] = 1
n.synapses['delay'][1] = 0
n.synapses['delay'][14] = 0
# Train neuron with 2 patterns
wA = Word([(1,1), (6,0), (9,0), (14,1)])
wB = Word([(3,0), (4,0), (9,0), (13,0)])
n.start_training()
assert_true(n.train(wA))
assert_true(n.train(wB))
n.finish_training()
# Recognize
wD = Word([(2,0), (6,0), (12,0), (14,0)])
fired, delay, container = n.expose(wD)
assert_false(fired)
wE = Word([(1,1), (3,0), (9,0), (13,0)])
fired, delay, container = n.expose(wE)
assert_true(fired) # False alarm
wF = Word([(1,0), (4,1), (7,0), (9,0), (11,0), (14,0)])
fired, delay, container = n.expose(wF)
assert_false(fired)
def test_train(self):
n = Neuron(S0 = 16, H = 4.0, G = 2.0, C = 1, D1 = 1, D2 = 1)
# Train neuron with 2 patterns
wA = Word([(1,0), (6,0), (9,0), (14,0)])
wB = Word([(3,0), (4,0), (9,0), (13,0)])
n.start_training()
assert_true(n.train(wA))
assert_true(n.train(wB))
n.finish_training()
# Test recognition
wD = Word([(2,0), (6,0), (12,0), (14,0)])
fired, delay, container = n.expose(wD)
assert_false(fired)
wE = Word([(3,0), (7,0), (9,0), (13,0)])
fired, delay, container = n.expose(wE)
assert_false(fired)
wF = Word([(1,0), (4,0), (9,0), (14,0)])
fired, delay, container = n.expose(wF)
assert_true(fired) # False alarm
