改变自训练阈值的影响#

此示例说明了改变自训练阈值的影响。加载 breast_cancer 数据集，并删除标签，以便只有 50 个样本中有 569 个样本有标签。在该数据集上拟合 SelfTrainingClassifier，并使用不同的阈值。

上图显示了分类器在拟合结束时可用的标记样本数量和分类器的准确率。下图显示了标记样本的最后一次迭代。所有值都使用 3 折交叉验证。

在低阈值（在 [0.4, 0.5] 中）时，分类器从使用低置信度标记的样本中学习。这些低置信度样本很可能具有错误的预测标签，因此，在这些错误标签上拟合会导致较差的准确率。请注意，分类器几乎标记了所有样本，并且只进行了一次迭代。

对于非常高的阈值（在 [0.9, 1) 之间），我们观察到分类器不会扩充其数据集（自标记样本的数量为 0）。因此，使用 0.9999 的阈值达成的准确率与普通监督分类器达成的准确率相同。

最佳准确率介于这两个极端之间，阈值约为 0.7。

# Authors: Oliver Rausch <[email protected]>
# License: BSD

import matplotlib.pyplot as plt
import numpy as np

from sklearn import datasets
from sklearn.metrics import accuracy_score
from sklearn.model_selection import StratifiedKFold
from sklearn.semi_supervised import SelfTrainingClassifier
from sklearn.svm import SVC
from sklearn.utils import shuffle

n_splits = 3

X, y = datasets.load_breast_cancer(return_X_y=True)
X, y = shuffle(X, y, random_state=42)
y_true = y.copy()
y[50:] = -1
total_samples = y.shape[0]

base_classifier = SVC(probability=True, gamma=0.001, random_state=42)

x_values = np.arange(0.4, 1.05, 0.05)
x_values = np.append(x_values, 0.99999)
scores = np.empty((x_values.shape[0], n_splits))
amount_labeled = np.empty((x_values.shape[0], n_splits))
amount_iterations = np.empty((x_values.shape[0], n_splits))

for i, threshold in enumerate(x_values):
    self_training_clf = SelfTrainingClassifier(base_classifier, threshold=threshold)

    # We need manual cross validation so that we don't treat -1 as a separate
    # class when computing accuracy
    skfolds = StratifiedKFold(n_splits=n_splits)
    for fold, (train_index, test_index) in enumerate(skfolds.split(X, y)):
        X_train = X[train_index]
        y_train = y[train_index]
        X_test = X[test_index]
        y_test = y[test_index]
        y_test_true = y_true[test_index]

        self_training_clf.fit(X_train, y_train)

        # The amount of labeled samples that at the end of fitting
        amount_labeled[i, fold] = (
            total_samples
            - np.unique(self_training_clf.labeled_iter_, return_counts=True)[1][0]
        )
        # The last iteration the classifier labeled a sample in
        amount_iterations[i, fold] = np.max(self_training_clf.labeled_iter_)

        y_pred = self_training_clf.predict(X_test)
        scores[i, fold] = accuracy_score(y_test_true, y_pred)


ax1 = plt.subplot(211)
ax1.errorbar(
    x_values, scores.mean(axis=1), yerr=scores.std(axis=1), capsize=2, color="b"
)
ax1.set_ylabel("Accuracy", color="b")
ax1.tick_params("y", colors="b")

ax2 = ax1.twinx()
ax2.errorbar(
    x_values,
    amount_labeled.mean(axis=1),
    yerr=amount_labeled.std(axis=1),
    capsize=2,
    color="g",
)
ax2.set_ylim(bottom=0)
ax2.set_ylabel("Amount of labeled samples", color="g")
ax2.tick_params("y", colors="g")

ax3 = plt.subplot(212, sharex=ax1)
ax3.errorbar(
    x_values,
    amount_iterations.mean(axis=1),
    yerr=amount_iterations.std(axis=1),
    capsize=2,
    color="b",
)
ax3.set_ylim(bottom=0)
ax3.set_ylabel("Amount of iterations")
ax3.set_xlabel("Threshold")

plt.show()

脚本总运行时间：（0 分钟 5.548 秒）