-
Notifications
You must be signed in to change notification settings - Fork 5
/
plot_tool.py
172 lines (131 loc) · 4.33 KB
/
plot_tool.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
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
import numpy as np
import matplotlib.pyplot as plt
from scipy.stats import multivariate_normal
from sklearn.manifold import TSNE
def sample_gaussian(dist, ns=1000):
m, ls = np.split(dist, 2, axis=-1)
s = np.exp(ls / 2)
s = np.diag(s)
return np.random.multivariate_normal(m, s,ns)
def sample_mgaussian(dists, mixw, ns=3000):
num_mode = len(mixw)
all_points = []
for n in range(ns):
di=np.random.choice(num_mode, 1, p=mixw)[0]
p = sample_gaussian(dists[di], ns=1)[0]
all_points.append(p)
return all_points
def plot_tsne(dist1, mixw, dist2):
point1s = sample_mgaussian(dist1, mixw, ns=100)
point2s = sample_gaussian(dist2, ns=100)
tsne_model = TSNE(perplexity=40, n_components=2, init='pca', n_iter=500, random_state=23)
new_value1s = tsne_model.fit_transform(point1s)
new_value2s = tsne_model.fit_transform(point2s)
x1 = []
y1 = []
for value in new_value1s:
x1.append(value[0])
y1.append(value[1])
x2 = []
y2 = []
for value in new_value2s:
x2.append(value[0])
y2.append(value[1])
fig, ax = plt.subplots(nrows=1, ncols=2, figsize=(6, 6))
maxx = 0
maxy = 0
for i in range(len(x1)):
ax[0].scatter(x1[i], y1[i], c='b')
maxx = max(maxx, abs(x1[i]))
maxy = max(maxy, abs(y1[i]))
ax[0].set_xlim(-maxx*2, maxx*2)
ax[0].set_ylim(-maxy*2, maxy*2)
maxx = 0
maxy = 0
for i in range(len(x2)):
ax[1].scatter(x2[i], y2[i], c='r')
maxx = max(maxx, abs(x2[i]))
maxy = max(maxy, abs(y2[i]))
ax[1].set_xlim(-maxx * 2, maxx * 2)
ax[1].set_ylim(-maxy * 2, maxy * 2)
plt.show()
def plot_tsne2(dist1s, mixws, dist2s, dstr):
seq_len = len(dist1s)
fig, ax = plt.subplots(nrows=2, ncols=seq_len)
c=0
ldstr = dstr.strip().split()
for dist1,mixw, dist2 in zip(dist1s, mixws, dist2s):
point1s = sample_mgaussian(dist1, mixw, ns=100)
point2s = sample_gaussian(dist2, ns=100)
tsne_model = TSNE(perplexity=40, n_components=2, init='pca', n_iter=500, random_state=23)
new_value1s = tsne_model.fit_transform(point1s)
new_value2s = tsne_model.fit_transform(point2s)
x1 = []
y1 = []
for value in new_value1s:
x1.append(value[0])
y1.append(value[1])
x2 = []
y2 = []
for value in new_value2s:
x2.append(value[0])
y2.append(value[1])
maxx = 0
maxy = 0
for i in range(len(x1)):
ax[0][c].scatter(x1[i], y1[i], c='b')
maxx = max(maxx, abs(x1[i]))
maxy = max(maxy, abs(y1[i]))
ax[0][c].set_title(ldstr[c])
ax[0][c].set_xlim(-maxx*2, maxx*2)
ax[0][c].set_ylim(-maxy*2, maxy*2)
maxx = 0
maxy = 0
for i in range(len(x2)):
ax[1][c].scatter(x2[i], y2[i], c='r')
maxx = max(maxx, abs(x2[i]))
maxy = max(maxy, abs(y2[i]))
ax[1][c].set_xlim(-maxx * 2, maxx * 2)
ax[1][c].set_ylim(-maxy * 2, maxy * 2)
c+=1
fig.tight_layout()
plt.show()
def plot_mgauss(dist1, mixw, dist2, xlim = (-5, 5), ylim = (-5, 5), xres = 500, yres = 500):
x = np.linspace(xlim[0], xlim[1], xres)
y = np.linspace(ylim[0], ylim[1], yres)
xx, yy = np.meshgrid(x, y)
# evaluate kernels at grid points
xxyy = np.c_[xx.ravel(), yy.ravel()]
zz = None
# print(dist1.shape)
for di, d in enumerate(dist1):
m, ls = np.split(d, 2, axis=-1)
m = m[-2:]
ls = ls[-2:]
s = np.exp(ls / 2)
s = np.diag(s)
k = multivariate_normal(mean=m, cov=s)
# print(m)
# print(s)
# print(mixw[di])
# print('--')
if di==0:
zz = mixw[di]*k.pdf(xxyy)
else:
zz += mixw[di]*k.pdf(xxyy)
# reshape and plot image
img = zz.reshape((xres,yres))
m, ls = np.split(dist2, 2, axis=-1)
m = m[-2:]
ls = ls[-2:]
s = np.exp(ls / 2)
s = np.diag(s)
k = multivariate_normal(mean=m, cov=s)
zz2 = k.pdf(xxyy)
img2 = zz2.reshape((xres, yres))
fig, ax = plt.subplots(nrows=1, ncols=2, figsize=(6, 6))
ax[0].imshow(img, cmap='Reds', extent=[-5, 5, -5, 5])
ax[1].imshow(img2, cmap='Reds', extent=[-5, 5, -5, 5])
plt.show()
if __name__ == '__main__':
plot_mgauss()