-
Notifications
You must be signed in to change notification settings - Fork 0
/
xor_investigate.py
194 lines (178 loc) · 6.18 KB
/
xor_investigate.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
import multiprocessing
import numpy as np
import tuning
import util
import xor
def xor_investigate(
seed=None,
dmg1=False,
dmg2=False,
te_mc=40.0,
te_mc_mu=5.0,
**kwargs):
#kwargs = {}
explorer = xor.solve_xor(
do_explore=False,
seed=seed,
**kwargs)
if False:
pivot = explorer.homomorphic_optimize(explorer.progenitor)
print("Pivot = %.2f" % pivot.fitness)
print(str(pivot.network))
max_mc = 20
attempts = 100
for mc in range(10, 51, 10):
successes = 0
scores = []
for attempt in range(attempts):
sp = explorer.explore_topology(pivot, mc)
spo = explorer.homomorphic_optimize(sp)
if spo.is_winner:
successes += 1
scores.append(spo.fitness)
print("%d %d" % (mc, successes))
#print(sorted(scores, reverse= True))
if True:
while True:
while True:
explorer.progenitor.explored_count = 0
pivot = explorer.homomorphic_optimize(explorer.progenitor, dmg=dmg1)
if pivot is not explorer.progenitor:
break
for i in range(10):
mc = explorer.random.gauss(te_mc, te_mc_mu)
mc = int(mc)
mc = max(mc, 1)
sp = explorer.explore_topology(pivot, mc)
spo = explorer.homomorphic_optimize(sp, dmg=dmg2)
if spo.is_winner:
steps = explorer.step
#print("Steps = %d" % steps)
return (steps, spo.get_size())
def xor_investigate_average():
n_runs = 100
x = []
for i in range(n_runs):
result =xor_investigate(i)
print("xor(%d) = %d" % (i, result))
x.append(result)
#x = [xor_investigate() for i in range(n_runs)]
print("Average=%.2f" % util.avg(x))
print("Median=%.2f" % util.median(x))
def xor_investigate_spread(kwargs, var, values, f=None, prefix=None):
kw2 = dict(kwargs)
kw2["func"] = xor_investigate
kw2["func_name"] = "Xor"
kw2["pool"] = pool
kw2["n_runs"] = n_runs
return tuning.func_spread(kw2, var, values, f, prefix)
# run1 yields 3600 on many values, but 3480 with mc_a=8
def run1():
kwargs = {}
#kwargs["dmg2"] = True
kwargs['max_steps'] = 20000
kwargs["start_with_connected_progenitor"] = True
kwargs["perturb_mutation_weight"] = 10
kwargs["pruning_discount_factor"] = 0.8
kwargs["dead_end_threshold"] = 1000
kwargs["subprogenitor_discount_factor"] = 0.9
kwargs['enable_homomorphic_propagation'] = False
#kwargs["mc_base"] = 1.2
kwargs["neuron_cost"] = 0.05
kwargs["fitness_threshold"] = 13.0
kwargs["pivotize_threshold"] = 20
kwargs["pivot_mutation_weights"] = [1,1,1]
f = open("xor_investigate.txt", "w")
r = range(1, 11)
xor_investigate_spread(kwargs, "mc_a", r, f=f)
# yields 3200 on mc_a=8 and te_mc = 25
def run2():
kwargs = {}
#kwargs["dmg2"] = True
kwargs["te_mc"] = 25
kwargs['max_steps'] = 20000
kwargs["start_with_connected_progenitor"] = True
kwargs["perturb_mutation_weight"] = 10
kwargs["pruning_discount_factor"] = 0.8
kwargs["dead_end_threshold"] = 1000
kwargs["subprogenitor_discount_factor"] = 0.9
kwargs['enable_homomorphic_propagation'] = False
#kwargs["mc_base"] = 1.2
kwargs["neuron_cost"] = 0.05
kwargs["fitness_threshold"] = 13.0
kwargs["pivotize_threshold"] = 20
kwargs["pivot_mutation_weights"] = [1,1,1]
f = open("xor_investigate.txt", "w")
r = [1,6,8]
xor_investigate_spread(kwargs, "mc_a", r, f=f)
class MyConstantMcClass(object):
def __init__(self, i):
self.i = i
def __call__(self, org, random):
return self.i
def __str__(self):
return "const_mc_%d" % self.i
#def my_mc_func(org, random):
# return mc
def constant_mc_func(mc):
return MyConstantMcClass(mc)
class MyProportionalMcClass(object):
def __init__(self, p):
self.p = p
def __call__(self, org, random):
network = org.network
connections = sum([len(n.connections) + len(n.incoming_connections) for n in network.neurons.values()]) / 2
#print(connections)
result = int(self.p * connections)
result = max(result, 2)
#print(result)
return result
def __str__(self):
return "proportional_mc_%.2f" % self.p
def run3():
kwargs = {}
#kwargs["dmg2"] = True
kwargs["te_mc"] = 25
kwargs['max_steps'] = 20000
kwargs["start_with_connected_progenitor"] = True
kwargs["perturb_mutation_weight"] = 10
kwargs["pruning_discount_factor"] = 0.8
kwargs["dead_end_threshold"] = 1000
kwargs["subprogenitor_discount_factor"] = 0.9
kwargs['enable_homomorphic_propagation'] = False
#kwargs["mc_base"] = 1.2
kwargs["neuron_cost"] = 0.05
kwargs["fitness_threshold"] = 13.0
kwargs["pivotize_threshold"] = 20
kwargs["pivot_mutation_weights"] = [1,1,1]
kwargs["seed_base"] = 1000
f = open("xor_investigate.txt", "w")
r = map(constant_mc_func, range(3,4,1))
xor_investigate_spread(kwargs, "mc_func", r, f=f)
#kwargs["seed_base"] = 100
#xor_investigate_spread(kwargs, "mc_func", r, f=f,prefix="sb100")
def run4():
kwargs = {}
#kwargs["dmg2"] = True
kwargs["te_mc"] = 25
kwargs['max_steps'] = 20000
kwargs["start_with_connected_progenitor"] = True
kwargs["perturb_mutation_weight"] = 10
kwargs["pruning_discount_factor"] = 0.8
kwargs["dead_end_threshold"] = 1000
kwargs["subprogenitor_discount_factor"] = 0.9
kwargs['enable_homomorphic_propagation'] = False
#kwargs["mc_base"] = 1.2
kwargs["neuron_cost"] = 0.05
kwargs["fitness_threshold"] = 13.0
kwargs["pivotize_threshold"] = 20
kwargs["pivot_mutation_weights"] = [1,1,1]
f = open("xor_investigate.txt", "w")
r = map(MyProportionalMcClass, np.arange(.05, .16, .025))
xor_investigate_spread(kwargs, "mc_func", r, f=f)
pool = None
n_runs =10
if __name__ == '__main__':
multiprocessing .freeze_support()
pool = multiprocessing.Pool(3)
run3()