|
1
|
|
|
from hyperactive.opt._adapters._gfo import _BaseGFOadapter |
|
2
|
|
|
|
|
3
|
|
|
|
|
4
|
|
|
class SimulatedAnnealing(_BaseGFOadapter): |
|
5
|
|
|
"""Simulated annealing optimizer. |
|
6
|
|
|
|
|
7
|
|
|
Parameters |
|
8
|
|
|
---------- |
|
9
|
|
|
search_space : dict[str, list] |
|
10
|
|
|
The search space to explore. A dictionary with parameter |
|
11
|
|
|
names as keys and a numpy array as values. |
|
12
|
|
|
initialize : dict[str, int] |
|
13
|
|
|
The method to generate initial positions. A dictionary with |
|
14
|
|
|
the following key literals and the corresponding value type: |
|
15
|
|
|
{"grid": int, "vertices": int, "random": int, "warm_start": list[dict]} |
|
16
|
|
|
constraints : list[callable] |
|
17
|
|
|
A list of constraints, where each constraint is a callable. |
|
18
|
|
|
The callable returns `True` or `False` dependend on the input parameters. |
|
19
|
|
|
random_state : None, int |
|
20
|
|
|
If None, create a new random state. If int, create a new random state |
|
21
|
|
|
seeded with the value. |
|
22
|
|
|
rand_rest_p : float |
|
23
|
|
|
The probability of a random iteration during the the search process. |
|
24
|
|
|
epsilon : float |
|
25
|
|
|
The step-size for the climbing. |
|
26
|
|
|
distribution : str |
|
27
|
|
|
The type of distribution to sample from. |
|
28
|
|
|
n_neighbours : int |
|
29
|
|
|
The number of neighbours to sample and evaluate before moving to the best |
|
30
|
|
|
of those neighbours. |
|
31
|
|
|
annealing_rate : float |
|
32
|
|
|
The rate at which the temperature is annealed. |
|
33
|
|
|
start_temp : float |
|
34
|
|
|
The initial temperature. |
|
35
|
|
|
n_iter : int, default=100 |
|
36
|
|
|
The number of iterations to run the optimizer. |
|
37
|
|
|
verbose : bool, default=False |
|
38
|
|
|
If True, print the progress of the optimization process. |
|
39
|
|
|
experiment : BaseExperiment, optional |
|
40
|
|
|
The experiment to optimize parameters for. |
|
41
|
|
|
Optional, can be passed later via ``set_params``. |
|
42
|
|
|
|
|
43
|
|
|
Examples |
|
44
|
|
|
-------- |
|
45
|
|
|
Basic usage of SimulatedAnnealing with a scikit-learn experiment: |
|
46
|
|
|
|
|
47
|
|
|
1. defining the experiment to optimize: |
|
48
|
|
|
>>> from hyperactive.experiment.integrations import SklearnCvExperiment |
|
49
|
|
|
>>> from sklearn.datasets import load_iris |
|
50
|
|
|
>>> from sklearn.svm import SVC |
|
51
|
|
|
>>> |
|
52
|
|
|
>>> X, y = load_iris(return_X_y=True) |
|
53
|
|
|
>>> |
|
54
|
|
|
>>> sklearn_exp = SklearnCvExperiment( |
|
55
|
|
|
... estimator=SVC(), |
|
56
|
|
|
... X=X, |
|
57
|
|
|
... y=y, |
|
58
|
|
|
... ) |
|
59
|
|
|
|
|
60
|
|
|
2. setting up the simulatedAnnealing optimizer: |
|
61
|
|
|
>>> from hyperactive.opt import SimulatedAnnealing |
|
62
|
|
|
>>> import numpy as np |
|
63
|
|
|
>>> |
|
64
|
|
|
>>> config = { |
|
65
|
|
|
... "search_space": { |
|
66
|
|
|
... "C": [0.01, 0.1, 1, 10], |
|
67
|
|
|
... "gamma": [0.0001, 0.01, 0.1, 1, 10], |
|
68
|
|
|
... }, |
|
69
|
|
|
... "n_iter": 100, |
|
70
|
|
|
... } |
|
71
|
|
|
>>> optimizer = SimulatedAnnealing(experiment=sklearn_exp, **config) |
|
72
|
|
|
|
|
73
|
|
|
3. running the optimization: |
|
74
|
|
|
>>> best_params = optimizer.run() |
|
75
|
|
|
|
|
76
|
|
|
Best parameters can also be accessed via: |
|
77
|
|
|
>>> best_params = optimizer.best_params_ |
|
78
|
|
|
""" |
|
79
|
|
|
|
|
80
|
|
|
_tags = { |
|
81
|
|
|
"info:name": "Simulated Annealing", |
|
82
|
|
|
"info:local_vs_global": "global", |
|
83
|
|
|
"info:explore_vs_exploit": "explore", |
|
84
|
|
|
"info:compute": "middle", |
|
85
|
|
|
} |
|
86
|
|
|
|
|
87
|
|
|
def __init__( |
|
88
|
|
|
self, |
|
89
|
|
|
search_space=None, |
|
90
|
|
|
initialize=None, |
|
91
|
|
|
constraints=None, |
|
92
|
|
|
random_state=None, |
|
93
|
|
|
rand_rest_p=0.1, |
|
94
|
|
|
epsilon=0.01, |
|
95
|
|
|
distribution="normal", |
|
96
|
|
|
n_neighbours=10, |
|
97
|
|
|
annealing_rate=0.97, |
|
98
|
|
|
start_temp=1, |
|
99
|
|
|
n_iter=100, |
|
100
|
|
|
verbose=False, |
|
101
|
|
|
experiment=None, |
|
102
|
|
|
): |
|
103
|
|
|
self.random_state = random_state |
|
104
|
|
|
self.rand_rest_p = rand_rest_p |
|
105
|
|
|
self.epsilon = epsilon |
|
106
|
|
|
self.distribution = distribution |
|
107
|
|
|
self.n_neighbours = n_neighbours |
|
108
|
|
|
self.annealing_rate = annealing_rate |
|
109
|
|
|
self.start_temp = start_temp |
|
110
|
|
|
self.search_space = search_space |
|
111
|
|
|
self.initialize = initialize |
|
112
|
|
|
self.constraints = constraints |
|
113
|
|
|
self.n_iter = n_iter |
|
114
|
|
|
self.experiment = experiment |
|
115
|
|
|
self.verbose = verbose |
|
116
|
|
|
|
|
117
|
|
|
super().__init__() |
|
118
|
|
|
|
|
119
|
|
|
def _get_gfo_class(self): |
|
120
|
|
|
"""Get the GFO class to use. |
|
121
|
|
|
|
|
122
|
|
|
Returns |
|
123
|
|
|
------- |
|
124
|
|
|
class |
|
125
|
|
|
The GFO class to use. One of the concrete GFO classes |
|
126
|
|
|
""" |
|
127
|
|
|
from gradient_free_optimizers import SimulatedAnnealingOptimizer |
|
128
|
|
|
|
|
129
|
|
|
return SimulatedAnnealingOptimizer |
|
130
|
|
|
|
|
131
|
|
|
@classmethod |
|
132
|
|
|
def get_test_params(cls, parameter_set="default"): |
|
133
|
|
|
"""Get the test parameters for the optimizer. |
|
134
|
|
|
|
|
135
|
|
|
Returns |
|
136
|
|
|
------- |
|
137
|
|
|
dict with str keys |
|
138
|
|
|
The test parameters dictionary. |
|
139
|
|
|
""" |
|
140
|
|
|
import numpy as np |
|
141
|
|
|
|
|
142
|
|
|
params = super().get_test_params() |
|
143
|
|
|
experiment = params[0]["experiment"] |
|
144
|
|
|
more_params = { |
|
145
|
|
|
"experiment": experiment, |
|
146
|
|
|
"start_temp": 0.33, |
|
147
|
|
|
"annealing_rate": 1.01, |
|
148
|
|
|
"search_space": { |
|
149
|
|
|
"C": [0.01, 0.1, 1, 10], |
|
150
|
|
|
"gamma": [0.0001, 0.01, 0.1, 1, 10], |
|
151
|
|
|
}, |
|
152
|
|
|
"n_iter": 100, |
|
153
|
|
|
} |
|
154
|
|
|
params.append(more_params) |
|
155
|
|
|
return params |
|
156
|
|
|
|
SimonBlanke /
Hyperactive
