|
1
|
|
|
# Author: Simon Blanke |
|
2
|
|
|
# Email: [email protected] |
|
3
|
|
|
# License: MIT License |
|
4
|
|
|
|
|
5
|
|
|
import numpy as np |
|
6
|
|
|
|
|
7
|
|
|
from sklearn.datasets import load_iris |
|
8
|
|
|
from sklearn.model_selection import cross_val_score |
|
9
|
|
|
from sklearn.tree import DecisionTreeClassifier |
|
10
|
|
|
from hyperactive import Hyperactive |
|
11
|
|
|
|
|
12
|
|
|
data = load_iris() |
|
13
|
|
|
X = data.data |
|
14
|
|
|
y = data.target |
|
15
|
|
|
|
|
16
|
|
|
memory = False |
|
17
|
|
|
|
|
18
|
|
|
n_iter = 100 |
|
19
|
|
|
|
|
20
|
|
|
|
|
21
|
|
|
def sphere_function(para, X_train, y_train): |
|
22
|
|
|
loss = [] |
|
23
|
|
|
for key in para.keys(): |
|
24
|
|
|
if key == "iteration": |
|
25
|
|
|
continue |
|
26
|
|
|
loss.append(para[key] * para[key]) |
|
27
|
|
|
|
|
28
|
|
|
return -np.array(loss).sum() |
|
29
|
|
|
|
|
30
|
|
|
|
|
31
|
|
|
search_config = {sphere_function: {"x1": range(-10, 10), "x2": range(-10, 10)}} |
|
32
|
|
|
|
|
33
|
|
|
|
|
34
|
|
|
def test_HillClimbingOptimizer(): |
|
35
|
|
|
opt = Hyperactive(X, y, memory=memory) |
|
36
|
|
|
opt.search(search_config, n_iter=n_iter, optimizer="HillClimbing") |
|
37
|
|
|
|
|
38
|
|
|
|
|
39
|
|
|
def test_StochasticHillClimbingOptimizer(): |
|
40
|
|
|
opt = Hyperactive(X, y, memory=memory) |
|
41
|
|
|
opt.search(search_config, n_iter=n_iter, optimizer="StochasticHillClimbing") |
|
42
|
|
|
|
|
43
|
|
|
|
|
44
|
|
|
def test_TabuOptimizer(): |
|
45
|
|
|
opt = Hyperactive(X, y, memory=memory) |
|
46
|
|
|
opt.search(search_config, n_iter=n_iter, optimizer="TabuSearch") |
|
47
|
|
|
|
|
48
|
|
|
|
|
49
|
|
|
def test_RandomSearchOptimizer(): |
|
50
|
|
|
opt = Hyperactive(X, y, memory=memory) |
|
51
|
|
|
opt.search(search_config, n_iter=n_iter, optimizer="RandomSearch") |
|
52
|
|
|
|
|
53
|
|
|
|
|
54
|
|
|
def test_RandomRestartHillClimbingOptimizer(): |
|
55
|
|
|
opt = Hyperactive(X, y, memory=memory) |
|
56
|
|
|
opt.search(search_config, n_iter=n_iter, optimizer="RandomRestartHillClimbing") |
|
57
|
|
|
|
|
58
|
|
|
|
|
59
|
|
|
def test_RandomAnnealingOptimizer(): |
|
60
|
|
|
opt = Hyperactive(X, y, memory=memory) |
|
61
|
|
|
opt.search(search_config, n_iter=n_iter, optimizer="RandomAnnealing") |
|
62
|
|
|
|
|
63
|
|
|
|
|
64
|
|
|
def test_SimulatedAnnealingOptimizer(): |
|
65
|
|
|
opt = Hyperactive(X, y, memory=memory) |
|
66
|
|
|
opt.search(search_config, n_iter=n_iter, optimizer="SimulatedAnnealing") |
|
67
|
|
|
|
|
68
|
|
|
|
|
69
|
|
|
def test_StochasticTunnelingOptimizer(): |
|
70
|
|
|
opt = Hyperactive(X, y, memory=memory) |
|
71
|
|
|
opt.search(search_config, n_iter=n_iter, optimizer="StochasticTunneling") |
|
72
|
|
|
|
|
73
|
|
|
|
|
74
|
|
|
def test_ParallelTemperingOptimizer(): |
|
75
|
|
|
opt = Hyperactive(X, y, memory=memory) |
|
76
|
|
|
opt.search(search_config, n_iter=n_iter, optimizer="ParallelTempering") |
|
77
|
|
|
|
|
78
|
|
|
|
|
79
|
|
|
def test_ParticleSwarmOptimizer(): |
|
80
|
|
|
opt = Hyperactive(X, y, memory=memory) |
|
81
|
|
|
opt.search(search_config, n_iter=n_iter, optimizer="ParticleSwarm") |
|
82
|
|
|
|
|
83
|
|
|
|
|
84
|
|
|
def test_EvolutionStrategyOptimizer(): |
|
85
|
|
|
opt = Hyperactive(X, y, memory=memory) |
|
86
|
|
|
opt.search(search_config, n_iter=n_iter, optimizer="EvolutionStrategy") |
|
87
|
|
|
|
|
88
|
|
|
|
|
89
|
|
|
def test_BayesianOptimizer(): |
|
90
|
|
|
opt = Hyperactive(X, y, memory=memory) |
|
91
|
|
|
opt.search(search_config, n_iter=int(n_iter / 10), optimizer="Bayesian") |
|
92
|
|
|
|
SimonBlanke /
Hyperactive
