def test_excercise_coursera(self):
"""
x1 = 2, w1 = 1
x2 = 1, w2 = 0.5
x3 = 1, w3 = 0
bias = 0.5
sum(xi*wi) + bias = 3
Linear Model => output=3
Binary Threshold Model => output=1
Rectified Linear Model => output=3
Segmoid Model => output= 0.95
"""
#- Linear -#
n1 = Neuron(value=2, weight=1, model=Neuron.LINEAR)
n2 = Neuron(value=1, weight=0.5, model=Neuron.LINEAR)
n3 = Neuron(value=1, weight=0, model=Neuron.LINEAR)
linear_model = Neuron(bias=0.5, model=Neuron.LINEAR, neuron_input_list=[n1, n2, n3])
self.assertEquals(linear_model.fire(), 3)
#- Binary Threshold -#
n1 = Neuron(value=2, weight=1, model=Neuron.LINEAR)
n2 = Neuron(value=1, weight=0.5, model=Neuron.LINEAR)
n3 = Neuron(value=1, weight=0, model=Neuron.LINEAR)
linear_model = Neuron(bias=0.5, model=Neuron.BINARY_THRESHOLD, neuron_input_list=[n1, n2, n3])
self.assertEquals(linear_model.fire(), 1)
#- Rectified Linear -#
n1 = Neuron(value=2, weight=1, model=Neuron.LINEAR)
n2 = Neuron(value=1, weight=0.5, model=Neuron.LINEAR)
n3 = Neuron(value=1, weight=0, model=Neuron.LINEAR)
linear_model = Neuron(bias=0.5, model=Neuron.RECTIFIED_LINEAR, neuron_input_list=[n1, n2, n3])
self.assertEquals(linear_model.fire(), 3)
#- Segmoid -#
n1 = Neuron(value=2, weight=1, model=Neuron.LINEAR)
n2 = Neuron(value=1, weight=0.5, model=Neuron.LINEAR)
n3 = Neuron(value=1, weight=0, model=Neuron.LINEAR)
linear_model = Neuron(bias=0.5, model=Neuron.SEGMOID, neuron_input_list=[n1, n2, n3])
self.assertEquals(round(linear_model.fire(), 2), .95)
def test_segmoid(self):
n1 = Neuron(model=Neuron.SEGMOID)
self.assertEquals(n1.fire(), 0.5)
n2 = Neuron(value=200, model=Neuron.SEGMOID)
self.assertEquals(n2.fire(), 1)
n3 = Neuron(value=2000, model=Neuron.SEGMOID)
self.assertEquals(n3.fire(), 1)
n4 = Neuron(value=20000000000, model=Neuron.SEGMOID)
self.assertEquals(n4.fire(), 1)
n5 = Neuron(value=-200, model=Neuron.SEGMOID)
self.assertEquals(n5.fire(), 0)
n6 = Neuron(value=-2000, model=Neuron.SEGMOID)
self.assertEquals(n6.fire(), 0)
n7 = Neuron(value=-2000000, model=Neuron.SEGMOID)
self.assertEquals(n7.fire(), 0)
n8 = Neuron(value=-2000000, bias = 10, model=Neuron.SEGMOID)
self.assertEquals(n8.fire(), 0)
n9= Neuron(value=2000000, bias = 10, model=Neuron.SEGMOID)
self.assertEquals(n9.fire(), 1)
n10 = Neuron(value=5, bias=-5, model=Neuron.SEGMOID)
self.assertEquals(n10.fire(), 0.5)
n11 = Neuron(value=-200, bias=200, model=Neuron.SEGMOID)
self.assertEquals(n11.fire(), 0.5)
def test_linear_model(self):
n1 = Neuron(model=Neuron.LINEAR)
self.assertEquals(n1.fire(), 0)
n2 = Neuron(model=Neuron.LINEAR, value=0)
self.assertEquals(n2.fire(), 0)
n3 = Neuron(model=Neuron.LINEAR, value=1)
self.assertEquals(n3.fire(), 1)
n4 = Neuron(model=Neuron.LINEAR, value=3)
self.assertEquals(n4.fire(), 3)
n5 = Neuron(model=Neuron.LINEAR, value=-1)
self.assertEquals(n5.fire(), -1)
n6 = Neuron(model=Neuron.LINEAR, value=2, bias=3)
self.assertEquals(n6.fire(), 5)
def test_xor_gate(self):
input1 = Neuron(value=1, weight=1) #1
input2 = Neuron(value=1, weight=1) #1
input3 = Neuron(weight=1, threshold=0.5, neuron_input_list=[input1, input2])
input4 = Neuron(weight=-1, threshold=1.5, neuron_input_list=[input1, input2])
xor_gate = Neuron(threshold=0.5, neuron_input_list=[input3, input4])
self.assertEquals(xor_gate.fire(), 0)
input1 = Neuron(value=1, weight=1) #1
input2 = Neuron(weight=1) #0
input3 = Neuron(weight=1, threshold=0.5, neuron_input_list=[input1, input2])
input4 = Neuron(weight=-1, threshold=1.5, neuron_input_list=[input1, input2])
xor_gate = Neuron(threshold=0.5, neuron_input_list=[input3, input4])
self.assertEquals(xor_gate.fire(), 1)
input1 = Neuron(value=1, weight=1) #1
input2 = Neuron(weight=1) #0
input3 = Neuron(weight=1, threshold=0.5, neuron_input_list=[input2, input1])
input4 = Neuron(weight=-1, threshold=1.5, neuron_input_list=[input1, input2])
xor_gate = Neuron(threshold=0.5, neuron_input_list=[input4, input3])
self.assertEquals(xor_gate.fire(), 1)
input1 = Neuron(weight=1) #0
input2 = Neuron(weight=1) #0
input3 = Neuron(weight=1, threshold=0.5, neuron_input_list=[input1, input2])
input4 = Neuron(weight=-1, threshold=1.5, neuron_input_list=[input1, input2])
xor_gate = Neuron(threshold=0.5, neuron_input_list=[input3, input4])
self.assertEquals(xor_gate.fire(), 0)
#- XOR API -#
in1 = Neuron(value=1, weight=1) #1
in2 = Neuron(value=1, weight=1) #1
xor_gate = Neuron.xor_gate()
xor_gate.set_inputs([in1, in2])
self.assertEqual(xor_gate.fire(), 0)
in1 = Neuron(value=1, weight=1) #1
in2 = Neuron(weight=1) #0
xor_gate = Neuron.xor_gate()
xor_gate.set_inputs([in1, in2])
self.assertEqual(xor_gate.fire(), 1)
xor_gate.set_inputs([in2, in1])
self.assertEqual(xor_gate.fire(), 1)
xor_gate.set_inputs([in2, in2])
self.assertEqual(xor_gate.fire(), 0)
def test_or_gate(self):
input1 = Neuron(value=1, weight=1)
input2 = Neuron(value=1, weight=1)
or_gate = Neuron(threshold=.09, neuron_input_list=[input1, input2])
self.assertEqual(or_gate.fire(), 1) # True or True = True
input3 = Neuron(weight=1) # 0
input4 = Neuron(value=1, weight=1) # 1
or_gate2 = Neuron(threshold=.09, neuron_input_list=[input3, input4])
self.assertEqual(or_gate2.fire(), 1) # False or True = True
or_gate3 = Neuron(threshold=.09, neuron_input_list=[input4, input3])
self.assertEqual(or_gate3.fire(), 1) # True or False = True
input5 = Neuron(weight=1) # 0
input6 = Neuron(weight=1) # 0
or_gate4 = Neuron(threshold=.09, neuron_input_list=[input5, input6])
self.assertEqual(or_gate4.fire(), 0) # False or False = False
#- OR API -#
in1 = Neuron(value=1, weight=1) #1
in2 = Neuron(value=1, weight=1) #1
or_gate = Neuron.or_gate()
or_gate.set_inputs([in1, in2])
self.assertEqual(or_gate.fire(), 1)
in1 = Neuron(value=1, weight=1) #1
in2 = Neuron(weight=1) #0
or_gate = Neuron.or_gate()
or_gate.set_inputs([in1, in2])
self.assertEqual(or_gate.fire(), 1)
or_gate.set_inputs([in2, in1])
self.assertEqual(or_gate.fire(), 1)
or_gate.set_inputs([in2, in2])
self.assertEqual(or_gate.fire(), 0)
def test_and_gate(self):
input3 = Neuron(value=1, weight=1) #1
input4 = Neuron(value=1, weight=1) #1
self.assertEqual(input3.fire(), 1)
self.assertEqual(input4.fire(), 1)
and_gate = Neuron(threshold=1.5, neuron_input_list=[input3, input4])
self.assertEqual(and_gate.fire(), 1) # True & True = True
input5 = Neuron(value=0, weight=1) #0
input6 = Neuron(value=1, weight=1) #1
self.assertEqual(input5.fire(), 0)
self.assertEqual(input6.fire(), 1)
and_gate2 = Neuron(threshold=1.5, neuron_input_list=[input5, input6])
self.assertEqual(and_gate2.fire(), 0) # True & False = False
and_gate3 = Neuron(threshold=1.5, neuron_input_list=[input6, input5])
self.assertEqual(and_gate3.fire(), 0) # False & True = False
input7 = Neuron(weight=1) #0
input8 = Neuron(weight=1) #0
self.assertEqual(input7.fire(), 0)
self.assertEqual(input8.fire(), 0)
and_gate4 = Neuron(threshold=1.5, neuron_input_list=[input7, input8])
self.assertEqual(and_gate4.fire(), 0) # False & False = False
#- Test and API -#
in1 = Neuron(value=1, weight=1) #1
in2 = Neuron(value=1, weight=1) #1
and_gate = Neuron.and_gate()
and_gate.set_inputs([in1, in2])
self.assertEqual(and_gate.fire(), 1)
in1 = Neuron(value=1, weight=1) #1
in2 = Neuron(weight=1) #0
self.assertEquals(in1.fire(), 1)
self.assertEquals(in2.fire(), 0)
and_gate = Neuron.and_gate()
and_gate.set_inputs([in1, in2])
self.assertEqual(and_gate.fire(), 0)
and_gate.set_inputs([in2, in1])
self.assertEqual(and_gate.fire(), 0)
and_gate.set_inputs([in2, in2])
self.assertEqual(and_gate.fire(), 0)
def test_rectified_linear_model(self):
n1 = Neuron(model=Neuron.RECTIFIED_LINEAR)
self.assertEquals(n1.fire(), 0)
n2 = Neuron(value=1, model=Neuron.RECTIFIED_LINEAR)
self.assertEquals(n2.fire(), 1)
n3 = Neuron(value=2, model=Neuron.RECTIFIED_LINEAR)
self.assertEquals(n3.fire(), 2)
n4 = Neuron(value=10, model=Neuron.RECTIFIED_LINEAR)
self.assertEquals(n4.fire(), 10)
n5 = Neuron(value=-1, model=Neuron.RECTIFIED_LINEAR)
self.assertEquals(n5.fire(), 0)
n6 = Neuron(value=-10, bias=10, model=Neuron.RECTIFIED_LINEAR)
self.assertEquals(n6.fire(), 0)
n7 = Neuron(bias=10, value=-10, model=Neuron.RECTIFIED_LINEAR)
self.assertEquals(n7.fire(), 0)
n8 = Neuron(value=20, bias=-19, model=Neuron.RECTIFIED_LINEAR)
self.assertEquals(n8.fire(), 1)
def test_binary_threshold_model(self):
#- By Default we use Binary Threshold Model -#
#- No Bias provided, Value >= Threshold -> 1 else 0 -#
input1 = Neuron(value=1)
input2 = Neuron(value=1)
self.assertEqual(input1.fire(), 1)
self.assertEqual(input2.fire(), 1)
input1 = Neuron(value=0)
input2 = Neuron(value=0)
self.assertEqual(input1.fire(), 0)
self.assertEqual(input2.fire(), 0)
input1 = Neuron()
input2 = Neuron()
self.assertEqual(input1.fire(), 0)
self.assertEqual(input2.fire(), 0)
input1 = Neuron(value=2)
input2 = Neuron(value=2)
self.assertEqual(input1.fire(), 1)
self.assertEqual(input2.fire(), 1)
input1 = Neuron(value=-1)
input2 = Neuron(value=-1)
self.assertEqual(input1.fire(), 0)
self.assertEqual(input2.fire(), 0)
#- Bias provided, Value >= 0 -> 1 else 0 -#
input1 = Neuron(value=1, bias=1)
input2 = Neuron(value=1, bias=1)
self.assertEqual(input1.fire(), 1)
self.assertEqual(input2.fire(), 1)
input1 = Neuron(value=0, bias=1)
input2 = Neuron(value=0, bias=1)
self.assertEqual(input1.fire(), 1)
self.assertEqual(input2.fire(), 1)