/
figure_gen.py
74 lines (49 loc) · 1.4 KB
/
figure_gen.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
70
71
72
73
import GPy
import numpy as np
import matplotlib.pyplot as plt
X = np.array([1, 2, 2.2, 6]).reshape(-1, 1)
Y = np.array([0.3, 1., 1.1, 0.6]).reshape(-1, 1)
# Y += np.random.randn(*Y.shape) * 1e-6
# m = GPy.models.GPRegression(X, Y, normalizer=True)
# m.optimize_restarts(num_restarts=100)
# print(m.param_array)
plt.figure(figsize=(16, 5))
ax = plt.subplot(121)
m = GPy.models.GPRegression(X, Y)
m.optimize()
m.plot(ax=ax, legend=False)
ax = plt.subplot(122)
m = GPy.models.GPRegression(X, Y)
m.optimize()
m.Gaussian_noise.variance = 0.03
m.plot(ax=ax, legend=False)
plt.tight_layout()
plt.savefig("images/gp-noise.png")
### RBF params
plt.figure(figsize=(16, 5))
ax = plt.subplot(121)
m = GPy.models.GPRegression(X, Y)
m.optimize()
m.kern.lengthscale = 0.2
m.plot(ax=ax, legend=False)
ax = plt.subplot(122)
m = GPy.models.GPRegression(X, Y)
m.optimize()
m.kern.lengthscale = 2.
m.plot(ax=ax, legend=False)
plt.savefig("images/gp-lengthscale.png")
### Parameter optimization
X = np.array([1, 2, 2.2, 6]).reshape(-1, 1)
Y = np.array([0.3, 1., 1.2, 0.6]).reshape(-1, 1)
plt.figure(figsize=(16, 5))
ax = plt.subplot(121)
m = GPy.models.GPRegression(X, Y)
m.Gaussian_noise.variance = 0.
m.kern.lengthscale = 4.
m.kern.variance = .1
m.plot(ax=ax, legend=False)
ax = plt.subplot(122)
m = GPy.models.GPRegression(X, Y)
m.optimize()
m.plot(ax=ax, legend=False)
plt.savefig("images/gp-kernel-param-optimization.png")