-
Notifications
You must be signed in to change notification settings - Fork 41
/
Copy pathtest_instance_incremental.py
196 lines (147 loc) · 5.37 KB
/
test_instance_incremental.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
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
import numpy as np
import pytest
from continuum.datasets import InMemoryDataset
from continuum.scenarios import InstanceIncremental
# yapf: disable
@pytest.fixture
def numpy_data():
nb_classes = 6
nb_data = 100
x_train = []
y_train = []
for i in range(nb_classes):
x_train.append(np.ones((nb_data, 4, 4, 3), dtype=np.uint8) * i)
y_train.append(np.ones(nb_data) * i)
x_train = np.concatenate(x_train)
y_train = np.concatenate(y_train)
x_test = np.copy(x_train)
y_test = np.copy(y_train)
return (x_train, y_train.astype(int)), (x_test, y_test.astype(int))
@pytest.mark.parametrize("nb_tasks,nb_tasks_gt", [
(2, 2),
(6, 6),
(1, 1),
])
def test_instance_auto_nb_tasks(numpy_data, nb_tasks, nb_tasks_gt):
"""Test the InstanceIncremental loader when the dataset doesn't provide
any default number of tasks."""
train, test = numpy_data
dummy = InMemoryDataset(*train)
scenario = InstanceIncremental(dummy, nb_tasks=nb_tasks)
nb_classes = scenario.nb_classes
assert len(scenario) == nb_tasks_gt
for task_id, train_dataset in enumerate(scenario):
assert nb_classes == len(np.unique(train_dataset._y))
@pytest.fixture
def numpy_data_per_task():
nb_classes = 6
nb_tasks = 3
nb_data = 100
x_train = []
y_train = []
t_train = []
for i in range(nb_tasks):
for j in range(nb_classes):
x_train.append(np.ones((nb_data, 4, 4, 3), dtype=np.uint8) * i)
y_train.append(np.ones(nb_data) * j)
t_train.append(np.ones(nb_data) * i)
x_train = np.concatenate(x_train)
y_train = np.concatenate(y_train)
t_train = np.concatenate(t_train)
x_test = np.copy(x_train)
y_test = np.copy(y_train)
t_test = np.copy(t_train)
return (x_train, y_train.astype(int), t_train), (x_test, y_test.astype(int), t_test)
@pytest.mark.parametrize("nb_tasks,nb_tasks_gt", [
(2, 2),
(6, 6),
(None, 3),
])
def test_instance_default_nb_tasks(numpy_data_per_task, nb_tasks, nb_tasks_gt):
"""Test the InstanceIncremental loader when the dataset does provide
a default number of tasks."""
train, test = numpy_data_per_task
x_train, y_train, t_train = train
dummy = InMemoryDataset(x_train, y_train, t=t_train)
scenario = InstanceIncremental(dummy, nb_tasks=nb_tasks)
nb_classes = scenario.nb_classes
assert len(scenario) == nb_tasks_gt, (nb_tasks, nb_tasks_gt)
for task_id, train_dataset in enumerate(scenario):
assert nb_classes == len(np.unique(train_dataset._y))
unique_pixels = np.unique(train_dataset._x)
if nb_tasks is None:
assert len(unique_pixels) == 1 and unique_pixels[0] == float(task_id)
@pytest.mark.parametrize("nb_tasks,error", [(301, "error"), (300, None), (300, "warning")])
def test_too_many_tasks(numpy_data_per_task, nb_tasks, error):
train, test = numpy_data_per_task
x_train, y_train, t_train = train
if error == "warning":
x_train = x_train[:-50]
y_train = y_train[:-50]
t_train = t_train[:-50]
dummy = InMemoryDataset(x_train, y_train, t=t_train)
if error == "error":
with pytest.raises(Exception):
scenario = InstanceIncremental(dummy, nb_tasks=nb_tasks)
elif error == "warning":
with pytest.warns(Warning):
scenario = InstanceIncremental(dummy, nb_tasks=nb_tasks)
else:
scenario = InstanceIncremental(dummy, nb_tasks=nb_tasks)
@pytest.mark.parametrize("nb_tasks", [None, 0, -1])
def test_invalid(numpy_data_per_task, nb_tasks):
train, test = numpy_data_per_task
x_train, y_train, t_train = train
x_test, y_test, t_test = test
dummy = InMemoryDataset(x_train, y_train)
with pytest.raises(Exception):
InstanceIncremental(dummy, nb_tasks=nb_tasks)
@pytest.fixture
def equal_data():
x = np.ones((100, 4, 4, 3), dtype=np.uint8)
y = np.concatenate((
np.ones((25,), dtype=np.uint32) * 0,
np.ones((50,), dtype=np.uint32) * 1,
np.ones((10,), dtype=np.uint32) * 2,
np.ones((15,), dtype=np.uint32) * 3,
))
return x, y
def test_instance_data_split_equally(equal_data):
dataset = InMemoryDataset(*equal_data)
scenario = InstanceIncremental(dataset, nb_tasks=5)
c = 0
for taskset in scenario:
bincount = np.bincount(taskset._y)
assert bincount[0] == 5
assert bincount[1] == 10
assert bincount[2] == 2
assert bincount[3] == 3
c += len(taskset)
assert c == 100
@pytest.fixture
def unequal_data():
y = np.concatenate((
np.ones((22,), dtype=np.uint32) * 0,
np.ones((53,), dtype=np.uint32) * 1,
np.ones((19,), dtype=np.uint32) * 2,
np.ones((7,), dtype=np.uint32) * 3,
))
x = np.ones((len(y), 4, 4, 3), dtype=np.uint8)
return x, y
def test_instance_data_split_not_equally(unequal_data):
x, y = unequal_data
dataset = InMemoryDataset(x, y)
scenario = InstanceIncremental(dataset, nb_tasks=5)
c_0, c_1, c_2, c_3 = 0, 0, 0, 0
for taskset in scenario:
bincount = np.bincount(taskset._y)
c_0 += bincount[0]
c_1 += bincount[1]
c_2 += bincount[2]
c_3 += bincount[3]
bincount = np.bincount(y)
assert c_0 == bincount[0]
assert c_1 == bincount[1]
assert c_2 == bincount[2]
assert c_3 == bincount[3]
assert c_0 + c_1 + c_2 + c_3 == len(x)