|
1
|
|
|
# Author: Simon Blanke |
|
2
|
|
|
# Email: [email protected] |
|
3
|
|
|
# License: MIT License |
|
4
|
|
|
|
|
5
|
|
|
import time |
|
6
|
|
|
import warnings |
|
7
|
|
|
|
|
8
|
|
|
from .main_args import MainArgs |
|
9
|
|
|
from .opt_args import Arguments |
|
10
|
|
|
from .distribution import dist |
|
11
|
|
|
|
|
12
|
|
|
from .optimizers import ( |
|
13
|
|
|
HillClimbingOptimizer, |
|
14
|
|
|
StochasticHillClimbingOptimizer, |
|
15
|
|
|
TabuOptimizer, |
|
16
|
|
|
RandomSearchOptimizer, |
|
17
|
|
|
RandomRestartHillClimbingOptimizer, |
|
18
|
|
|
RandomAnnealingOptimizer, |
|
19
|
|
|
SimulatedAnnealingOptimizer, |
|
20
|
|
|
StochasticTunnelingOptimizer, |
|
21
|
|
|
ParallelTemperingOptimizer, |
|
22
|
|
|
ParticleSwarmOptimizer, |
|
23
|
|
|
EvolutionStrategyOptimizer, |
|
24
|
|
|
BayesianOptimizer, |
|
25
|
|
|
) |
|
26
|
|
|
|
|
27
|
|
|
|
|
28
|
|
|
def stop_warnings(): |
|
29
|
|
|
# because sklearn warnings are annoying when they appear 100 times |
|
30
|
|
|
def warn(*args, **kwargs): |
|
31
|
|
|
pass |
|
32
|
|
|
|
|
33
|
|
|
import warnings |
|
34
|
|
|
|
|
35
|
|
|
warnings.warn = warn |
|
36
|
|
|
|
|
37
|
|
|
|
|
38
|
|
|
|
|
39
|
|
|
class Hyperactive: |
|
40
|
|
|
def __init__( |
|
41
|
|
|
self, X, y, memory="long", random_state=False, verbosity=3, warnings=False |
|
42
|
|
|
): |
|
43
|
|
|
self.X = X |
|
44
|
|
|
self._main_args_ = MainArgs(X, y, memory, random_state, verbosity) |
|
45
|
|
|
|
|
46
|
|
|
if not warnings: |
|
47
|
|
|
stop_warnings() |
|
48
|
|
|
|
|
49
|
|
|
self.optimizer_dict = { |
|
50
|
|
|
"HillClimbing": HillClimbingOptimizer, |
|
51
|
|
|
"StochasticHillClimbing": StochasticHillClimbingOptimizer, |
|
52
|
|
|
"TabuSearch": TabuOptimizer, |
|
53
|
|
|
"RandomSearch": RandomSearchOptimizer, |
|
54
|
|
|
"RandomRestartHillClimbing": RandomRestartHillClimbingOptimizer, |
|
55
|
|
|
"RandomAnnealing": RandomAnnealingOptimizer, |
|
56
|
|
|
"SimulatedAnnealing": SimulatedAnnealingOptimizer, |
|
57
|
|
|
"StochasticTunneling": StochasticTunnelingOptimizer, |
|
58
|
|
|
"ParallelTempering": ParallelTemperingOptimizer, |
|
59
|
|
|
"ParticleSwarm": ParticleSwarmOptimizer, |
|
60
|
|
|
"EvolutionStrategy": EvolutionStrategyOptimizer, |
|
61
|
|
|
"Bayesian": BayesianOptimizer, |
|
62
|
|
|
} |
|
63
|
|
|
|
|
64
|
|
|
def search( |
|
65
|
|
|
self, |
|
66
|
|
|
search_config, |
|
67
|
|
|
n_iter=10, |
|
68
|
|
|
max_time=None, |
|
69
|
|
|
optimizer="RandomSearch", |
|
70
|
|
|
n_jobs=1, |
|
71
|
|
|
init_config=None, |
|
72
|
|
|
): |
|
73
|
|
|
|
|
74
|
|
|
self._main_args_.search_args( |
|
75
|
|
|
search_config, max_time, n_iter, optimizer, n_jobs, init_config |
|
76
|
|
|
) |
|
77
|
|
|
self._opt_args_ = Arguments(**self._main_args_.opt_para) |
|
78
|
|
|
optimizer_class = self.optimizer_dict[self._main_args_.optimizer] |
|
79
|
|
|
|
|
80
|
|
|
dist(optimizer_class, self._main_args_, self._opt_args_) |
|
81
|
|
|
|
|
82
|
|
|
|
|
83
|
|
|
|
SimonBlanke /
Hyperactive
