-
Notifications
You must be signed in to change notification settings - Fork 2
/
evaluate.py
61 lines (59 loc) · 2.57 KB
/
evaluate.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
N = 40
K = [5] #,4] #,3] #,4,5,6,7,8] #,9,10]
H = [1,2,3,4,5,6,7,8] #,32,128,512]
import itertools
import numpy
import random
from sklearn.neural_network import MLPClassifier
results = []
print("n","k","h","successful classifications", "rate")
for k in K:
numpy.random.seed(0)
# print data
for h in H:
numpy.random.seed(0)
for n in range(N):
n += 1
data_results = []
for r_data in range(20):
numpy.random.seed(r_data)
if False: #k == 1:
data = [[d] for d in range(N)]
else:
data = numpy.random.uniform(size=[N,k])
numpy.random.seed(0)
true_results = 0
if n <= 8:
labellist = ["".join(item) for item in itertools.product("10", repeat=n)]
else:
labellist = [bin(numpy.random.randint(2**(N+2)+1, 2**(N+2)+1+2**n))[-n:] for i in range(256)]
for labelstring in labellist:
labels = [int(i) for i in labelstring]
d = data[:n]
for r_mlp in range(10): #lbfgs
clf = MLPClassifier(
hidden_layer_sizes=(h,), random_state=r_mlp,
#activation='relu', solver="lbfgs",
activation='relu', solver="lbfgs",
alpha=0)
#clf = MLPClassifier(hidden_layer_sizes=(1,), random_state=r, activation='identity', solver="lbfgs")
clf.fit(d, labels)
if (clf.predict(d) == labels).all():
true_results += 1
break
# true_results are always an even number!
true_results += true_results % 2
if true_results == 2**min(n,8):
data_results.append(true_results)
break
if data_results and true_results > max(data_results):
print(n, k, h, true_results, true_results*1.0/2**min(n,8), "intermediate", r_data, max(data_results)*1.0/2**min(n,8))
data_results.append(true_results)
true_results = max(data_results)
print(n, k, h, true_results, true_results*1.0/2**min(n,8))
results.append((n, k, h, true_results, true_results*1.0/2**min(n,8)))
if true_results*1.0/2**min(n,8) < 0.5:
print "KVC(0.5): "+str((n-1,k, h))
print
break
print "done"