/
OneVsRest.php
256 lines (213 loc) · 6.38 KB
/
OneVsRest.php
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
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
<?php
namespace Rubix\ML\Classifiers;
use Rubix\ML\Learner;
use Rubix\ML\Parallel;
use Rubix\ML\Estimator;
use Rubix\ML\Persistable;
use Rubix\ML\Probabilistic;
use Rubix\ML\EstimatorType;
use Rubix\ML\Helpers\Params;
use Rubix\ML\Backends\Serial;
use Rubix\ML\Datasets\Dataset;
use Rubix\ML\Backends\Tasks\Proba;
use Rubix\ML\Traits\Multiprocessing;
use Rubix\ML\Traits\AutotrackRevisions;
use Rubix\ML\Backends\Tasks\TrainLearner;
use Rubix\ML\Specifications\DatasetIsLabeled;
use Rubix\ML\Specifications\DatasetIsNotEmpty;
use Rubix\ML\Specifications\SpecificationChain;
use Rubix\ML\Specifications\DatasetHasDimensionality;
use Rubix\ML\Specifications\LabelsAreCompatibleWithLearner;
use Rubix\ML\Specifications\SamplesAreCompatibleWithEstimator;
use Rubix\ML\Exceptions\InvalidArgumentException;
use Rubix\ML\Exceptions\RuntimeException;
use function Rubix\ML\array_transpose;
use function array_combine;
use function array_keys;
use function array_map;
use function array_sum;
/**
* One Vs Rest
*
* One Vs Rest is an ensemble learner that trains a binary classifier to predict a particular class
* vs every other class for every possible class. The final class prediction is the class whose
* binary classifier returned the highest probability. One of the features of One Vs Rest is that
* it allows you to build a multiclass classifier out of an ensemble of otherwise binary classifiers.
*
* @category Machine Learning
* @package Rubix/ML
* @author Andrew DalPino
*/
class OneVsRest implements Estimator, Learner, Probabilistic, Parallel, Persistable
{
use AutotrackRevisions, Multiprocessing;
/**
* The base classifier.
*
* @var Learner
*/
protected Learner $base;
/**
* A map of each class to its binary classifier.
*
* @var array<\Rubix\ML\Learner>
*/
protected array $classifiers = [
//
];
/**
* The dimensionality of the training set.
*
* @var int<0,max>|null
*/
protected ?int $featureCount = null;
/**
* @param Learner $base
* @throws InvalidArgumentException
*/
public function __construct(Learner $base)
{
if (!$base->type()->isClassifier()) {
throw new InvalidArgumentException('Base Learner must be'
. ' a classifier.');
}
if (!$base instanceof Probabilistic) {
throw new InvalidArgumentException('Base classifier must'
. ' implement the Probabilistic interface.');
}
$this->base = $base;
$this->backend = new Serial();
}
/**
* Return the estimator type.
*
* @internal
*
* @return EstimatorType
*/
public function type() : EstimatorType
{
return EstimatorType::classifier();
}
/**
* Return the data types that the estimator is compatible with.
*
* @internal
*
* @return list<\Rubix\ML\DataType>
*/
public function compatibility() : array
{
return $this->base->compatibility();
}
/**
* Return the settings of the hyper-parameters in an associative array.
*
* @internal
*
* @return mixed[]
*/
public function params() : array
{
return [
'base' => $this->base,
];
}
/**
* Has the learner been trained?
*
* @return bool
*/
public function trained() : bool
{
return !empty($this->classifiers);
}
/**
* Train the learner with a dataset.
*
* @param \Rubix\ML\Datasets\Labeled $dataset
*/
public function train(Dataset $dataset) : void
{
SpecificationChain::with([
new DatasetIsLabeled($dataset),
new DatasetIsNotEmpty($dataset),
new SamplesAreCompatibleWithEstimator($dataset, $this),
new LabelsAreCompatibleWithLearner($dataset, $this),
])->check();
$classes = $dataset->possibleOutcomes();
$this->backend->flush();
foreach ($classes as $class) {
$estimator = clone $this->base;
$subset = clone $dataset;
$binarize = function ($label) use ($class) {
return $label === $class ? 'y' : 'n';
};
$subset->transformLabels($binarize);
$task = new TrainLearner($estimator, $subset);
$this->backend->enqueue($task);
}
$classifiers = $this->backend->process();
$classifiers = array_combine($classes, $classifiers) ?: [];
$this->classifiers = $classifiers;
$this->featureCount = $dataset->numFeatures();
}
/**
* Make predictions from a dataset.
*
* @param Dataset $dataset
* @throws RuntimeException
* @return list<string>
*/
public function predict(Dataset $dataset) : array
{
return array_map('\Rubix\ML\argmax', $this->proba($dataset));
}
/**
* Estimate the joint probabilities for each possible outcome.
*
* @param Dataset $dataset
* @throws RuntimeException
* @return list<array<string,float>>
*/
public function proba(Dataset $dataset) : array
{
if (!$this->classifiers or !$this->featureCount) {
throw new RuntimeException('Estimator has not been trained.');
}
DatasetHasDimensionality::with($dataset, $this->featureCount)->check();
$this->backend->flush();
/** @var Probabilistic $estimator */
foreach ($this->classifiers as $estimator) {
$task = new Proba($estimator, $dataset);
$this->backend->enqueue($task);
}
$aggregate = $this->backend->process();
$aggregate = array_transpose($aggregate);
$classes = array_keys($this->classifiers);
$probabilities = [];
foreach ($aggregate as $votes) {
$dist = [];
foreach ($votes as $j => $proba) {
$dist[$classes[$j]] = $proba['y'];
}
$total = array_sum($dist);
foreach ($dist as &$probability) {
$probability /= $total;
}
$probabilities[] = $dist;
}
return $probabilities;
}
/**
* Return the string representation of the object.
*
* @internal
*
* @return string
*/
public function __toString() : string
{
return 'One Vs Rest (' . Params::stringify($this->params()) . ')';
}
}