qboost/demo.py at master · dwave-examples/qboost · GitHub

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
#    Copyright 2018 D-Wave Systems Inc.

#    Licensed under the Apache License, Version 2.0 (the "License")
#    you may not use this file except in compliance with the License.
#    You may obtain a copy of the License at

#        http: // www.apache.org/licenses/LICENSE-2.0

#    Unless required by applicable law or agreed to in writing, software
#    distributed under the License is distributed on an "AS IS" BASIS,
#    WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
#    See the License for the specific language governing permissions and
#    limitations under the License.


import numpy as np
from sklearn.model_selection import train_test_split
from sklearn.datasets import load_digits
try:
    import matplotlib.pyplot as plt
except ImportError:
    # Not required for demo
    pass

from qboost import QBoostClassifier, qboost_lambda_sweep
from datasets import make_blob_data, get_handwritten_digits_data


if __name__ == '__main__':
    import argparse

    parser = argparse.ArgumentParser(description="Run QBoost example",
                                     epilog="Information about additional options that are specific to the data set can be obtained using either 'demo.py blobs -h' or 'demo.py digits -h'.")
    parser.add_argument('--verbose', action='store_true')
    parser.add_argument('--cross-validation', action='store_true',
                        help='use cross-validation to estimate the value of the regularization parameter')
    parser.add_argument('--lam', default=0.01, type=float,
                        help='regularization parameter (default: %(default)s)')

    # Note: required=True could be useful here, but not available
    # until Python 3.7
    subparsers = parser.add_subparsers(
        title='dataset', description='dataset to use', dest='dataset')

    sp_blobs = subparsers.add_parser('blobs', help='blobs data set')
    sp_blobs.add_argument('--num-samples', type=int, default=2000,
                          help='number of samples (default: %(default)s)')
    sp_blobs.add_argument('--num-features', type=int, default=10,
                          help='number of features (default: %(default)s)')
    sp_blobs.add_argument('--num-informative', type=int, default=2,
                          help='number of informative features (default: %(default)s)')

    sp_digits = subparsers.add_parser(
        'digits', help='handwritten digits data set')
    sp_digits.add_argument('--digit1', type=int, default=0, choices=range(10),
                           help='first digit to include (default: %(default)s)')
    sp_digits.add_argument('--digit2', type=int, default=1, choices=range(10),
                           help='second digit to include (default: %(default)s)')
    sp_digits.add_argument('--plot-digits', action='store_true',
                           help='plot a random sample of each digit')

    args = parser.parse_args()

    if args.dataset == 'blobs':
        n_samples = args.num_samples
        n_features = args.num_features
        n_informative = args.num_informative

        X, y = make_blob_data(
            n_samples=n_samples, n_features=n_features, n_informative=n_informative)

        X_train, X_test, y_train, y_test = train_test_split(
            X, y, test_size=0.4)

        if args.cross_validation:
            # See Boyda et al. (2017), Eq. (17) regarding normalization
            normalized_lambdas = np.linspace(0.0, 0.5, 10)
            lambdas = normalized_lambdas / n_features
            print('Performing cross-validation using {} values of lambda, this may take several minutes...'.format(len(lambdas)))
            qboost, lam = qboost_lambda_sweep(
                X_train, y_train, lambdas, verbose=args.verbose)
        else:
            qboost = QBoostClassifier(X_train, y_train, args.lam)

        if args.verbose:
            qboost.report_baseline(X_test, y_test)

        print('Informative features:', list(range(n_informative)))
        print('Selected features:', qboost.get_selected_features())

        print('Score on test set: {:.3f}'.format(qboost.score(X_test, y_test)))

    elif args.dataset == 'digits':
        if args.digit1 == args.digit2:
            raise ValueError("must use two different digits")

        X, y = get_handwritten_digits_data(args.digit1, args.digit2)
        n_features = np.size(X, 1)

        X_train, X_test, y_train, y_test = train_test_split(
            X, y, test_size=0.4)
        print('Number of features:', np.size(X, 1))
        print('Number of training samples:', len(X_train))
        print('Number of test samples:', len(X_test))

        if args.cross_validation:
            # See Boyda et al. (2017), Eq. (17) regarding normalization
            normalized_lambdas = np.linspace(0.0, 1.75, 10)
            lambdas = normalized_lambdas / n_features
            print('Performing cross-validation using {} values of lambda, this make take several minutes...'.format(len(lambdas)))
            qboost, lam = qboost_lambda_sweep(
                X_train, y_train, lambdas, verbose=args.verbose)
        else:
            qboost = QBoostClassifier(X_train, y_train, args.lam)

        if args.verbose:
            qboost.report_baseline(X_test, y_test)

        print('Number of selected features:',
              len(qboost.get_selected_features()))

        print('Score on test set: {:.3f}'.format(qboost.score(X_test, y_test)))

        if args.plot_digits:
            digits = load_digits()

            images1 = [image for image, target in zip(
                digits.images, digits.target) if target == args.digit1]
            images2 = [image for image, target in zip(
                digits.images, digits.target) if target == args.digit2]

            f, axes = plt.subplots(1, 2)

            # Select a random image from each set to show:
            i1 = np.random.choice(len(images1))
            i2 = np.random.choice(len(images2))
            for ax, image in zip(axes, (images1[i1], images2[i2])):
                ax.imshow(image, cmap=plt.cm.gray_r, interpolation='nearest')

            plt.show()

    elif not args.dataset:
        parser.print_help()