-
Notifications
You must be signed in to change notification settings - Fork 3
/
Copy pathdata_gen.py
179 lines (154 loc) · 7.33 KB
/
data_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
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
173
174
175
176
177
178
179
from pennylane import numpy as np
import pennylane as qml
import qiskit
import qiskit.providers.aer.noise as noise
import os
from tqdm import tqdm
import cv2
import math
import sys
sys.path.append('.')
from circuit import classifier, classifier_embedding
from tools.get_MNIST import load
def data_gen(n_qubits, root, n_depth_classifier=2, index=None, random_label=True, backend=None, net=None):
n_data = 1000
n_depth_embedding = 2
# if backend is None:
# dev = qml.device("default.qubit", wires=n_qubits)
# elif backend == 'qiskit.aer':
# # qiskit.IBMQ.load_account()
# provider = qiskit.IBMQ.providers(group='open')[0]
# backend_ibm = provider.get_backend('ibmq_16_melbourne') # ibmq_qasm_simulator
# noise_model = noise.NoiseModel.from_backend(backend_ibm)
# dev = qml.device('qiskit.aer', wires=n_qubits, noise_model=noise_model)
# else:
dev = qml.device("default.qubit", wires=n_qubits)
if net is None or net=='qnn':
circuit = qml.QNode(classifier, dev)
else:
circuit = qml.QNode(classifier_embedding, dev)
np.random.seed(index)
while True:
param = np.random.uniform(0, math.pi, n_qubits*n_depth_classifier*3)
param = np.reshape(param, (n_depth_classifier, n_qubits, 3))
data = np.random.uniform(0, 2*math.pi, n_data*n_depth_embedding*n_qubits)
data = np.array(data, requires_grad=False)
data = np.reshape(data, (n_data, n_depth_embedding, n_qubits))
gap_low, gap_high = 0.15, 0.15
data_save, label_save = [], []
for i in tqdm(range(n_data), desc='data_gen'):
exp = circuit(param, feat=data[i])
if exp < 0.0 - gap_low:
data_save.append(data[i])
label_save.append(-1)
elif exp > 0.0 + gap_high:
data_save.append(data[i])
label_save.append(1)
if sum(np.array(label_save) == 1) >= 200 and sum(np.array(label_save) == -1) >= 200:
break
print(sum(label_save))
label_save = np.array(label_save)
data_save = np.array(data_save)
dataset_type = '_true_'
if random_label:
np.random.seed(index)
random_index = np.random.permutation(len(label_save))
label_save = label_save[random_index]
dataset_type = '_random_'
np.save(os.path.join(root, 'feat'+dataset_type+net+'_'+str(n_qubits)+'_'+str(index)), data_save)
np.save(os.path.join(root, 'label'+dataset_type+net+'_'+str(n_qubits)+'_'+str(index)), label_save)
def get_data(feat_file=None, label_file=None, name='synthetic', random_label=True):
if name == 'synthetic':
feat = np.load(feat_file)
label = np.load(label_file)
index_neg = label==-1
index_pos = label==1
feat_pos_train = feat[index_pos][:100]
feat_pos_test = feat[index_pos][100:200]
feat_neg_train = feat[index_neg][:100]
feat_neg_test = feat[index_neg][100:200]
label_pos_train = label[index_pos][:100]
label_pos_test = label[index_pos][100:200]
label_neg_train = label[index_neg][:100]
label_neg_test = label[index_neg][100:200]
feat_train = np.array(np.concatenate((feat_pos_train, feat_neg_train), axis=0), requires_grad=False)
label_train = np.array(np.concatenate((label_pos_train, label_neg_train), axis=0), requires_grad=False)
feat_test = np.array(np.concatenate((feat_pos_test, feat_neg_test), axis=0), requires_grad=False)
label_test = np.array(np.concatenate((label_pos_test, label_neg_test), axis=0), requires_grad=False)
return feat_train, label_train, feat_test, label_test
elif name == 'MNIST':
feat_train, label_train, feat_test, label_test = load()
for i in range(len(feat_train)):
feat_train[i] = cv2.resize(feat_train[i], (10, 10), interpolation=cv2.INTER_NEAREST)
for i in range(len(feat_test)):
feat_test[i] = cv2.resize(feat_test[i], (10, 10), interpolation=cv2.INTER_NEAREST)
# transform data from [0, 255] to [-1, 1]
feat_train = feat_train / 255.0
feat_test = feat_test / 255.0
feat_train = math.pi*((feat_train - 0.1307) / 0.3081)
feat_test = math.pi*((feat_test - 0.1307) / 0.3081)
if random_label:
index = np.random.permutation(len(label_train))
label_train = label_train[index]
return feat_train, label_train, feat_test, label_test
elif name == 'WINE':
feat, label = [], []
with open(feat_file, 'r') as f:
for line in f.readlines():
parts = line.strip().split(',')
if float(parts[0]) > 2:
continue
label.append(2*(float(parts[0])-1)-1)
# label.append(float(parts[0])-1)
feat.append([float(part) for part in parts[1:]])
feat = np.array(feat, requires_grad=False)[:, np.newaxis, :]
# feat = 2*math.pi*(feat - np.min(feat, axis=0)) / np.ptp(feat, axis=0)
label = np.array(label, requires_grad=False)
if random_label:
index = np.random.permutation(len(label))
label = label[index]
index_neg = label==-1
index_pos = label==1
n_pos = sum(index_pos)
n_neg = sum(index_neg)
print('n_pos = {}, n_neg = {}'.format(n_pos, n_neg))
feat_pos_train = feat[index_pos][:int(0.5*n_pos)]
feat_pos_test = feat[index_pos][int(0.5*n_pos):]
feat_neg_train = feat[index_neg][:int(0.5*n_neg)]
feat_neg_test = feat[index_neg][int(0.5*n_neg):]
label_pos_train = label[index_pos][:int(0.5*n_pos)]
label_pos_test = label[index_pos][int(0.5*n_pos):]
label_neg_train = label[index_neg][:int(0.5*n_neg)]
label_neg_test = label[index_neg][int(0.5*n_neg):]
feat_train = np.array(np.concatenate((feat_pos_train, feat_neg_train), axis=0), requires_grad=False)
label_train = np.array(np.concatenate((label_pos_train, label_neg_train), axis=0), requires_grad=False)
feat_test = np.array(np.concatenate((feat_pos_test, feat_neg_test), axis=0), requires_grad=False)
label_test = np.array(np.concatenate((label_pos_test, label_neg_test), axis=0), requires_grad=False)
return feat_train, label_train, feat_test, label_test
if __name__ == '__main__':
# n_qubits = 10
# n_data = 1000
# n_depth_classifier = 2
# n_depth_embedding = 2
# dev = qml.device("default.qubit", wires=n_qubits)
# circuit = qml.QNode(classifier, dev)
# np.random.seed(0)
# param = np.random.uniform(0, math.pi, n_qubits*n_depth_classifier*3)
# param = np.reshape(param, (n_depth_classifier, n_qubits, 3))
# data = np.random.uniform(0, 2*math.pi, n_data*n_depth_embedding*n_qubits)
# data = np.array(data, requires_grad=False)
# data = np.reshape(data, (n_data, n_depth_embedding, n_qubits))
# gap_low, gap_high = 0.15, 0.15
# data_save, label_save = [], []
# for i in range(n_data):
# exp = circuit(param, data[i])
# if exp < 0.0 - gap_low:
# data_save.append(data[i])
# label_save.append(-1)
# elif exp > 0.0 + gap_high:
# data_save.append(data[i])
# label_save.append(1)
# print(sum(label_save))
# np.save('data/feat', np.array(data_save))
# np.save('data/label', np.array(label_save))
data_gen(16, 'data', 0)