-
Notifications
You must be signed in to change notification settings - Fork 2
/
exercise23.py
70 lines (50 loc) · 1.41 KB
/
exercise23.py
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
import numpy as np
from scipy.stats import norm
import matplotlib.pyplot as plt
rounds = 10000
ys = ([], [])
class postsynaptic_neuron:
def __init__(self):
self.omega = np.array([float(1),float(1)])
self.lrate = 0.005
self.theta = 100
self.delta = 2
self.realY = []
self.deltas = []
def bcm_rule(self, input):
y = np.inner(self.omega, input)
self.realY.append(y)
d_omega = self.lrate * input * (y**2 - y * self.delta)
self.omega += d_omega
self.omega[self.omega < 0] = 0
d_delta = (float(1) / self.theta * (y**2 - self.delta))
self.delta += d_delta
self.deltas.append(self.delta)
return self.omega
def get_parameters(self):
return [self.omega, self.deltas, self.realY]
input = [np.array([1,0]), np.array([0,1])]
postneuron = postsynaptic_neuron()
omega1 = []
omega2 = []
for i in range(0, rounds):
i = np.random.randint(0,2,1)
om = postneuron.bcm_rule(input[i])
print om[1]
omega1.append(om[0])
omega2.append(om[1])
# compute y for each of the five gaussians
[f_omega, f_deltas, f_y] = postneuron.get_parameters()
# compute the skewness for each iteration
fig = plt.figure()
fig.suptitle('Simulation for selectivity', fontsize=14, fontweight='bold')
fig.subplots_adjust(top=0.85, hspace=0.6)
ax1 = fig.add_subplot(311)
ax1.set_title('Weights evolution')
ax1.set_xlim(0, rounds)
ax1.plot(omega1)
ax1.plot(omega2)
ax1.plot(f_deltas)
# for i in range(0, 1):
# ax2.plot(ys[i])
fig.show()