度量学习(NCA)

from metric_learn import NCA

from sklearn.datasets import make_classification

from sklearn.neighbors import KNeighborsClassifier

nca = NCA()

X, y = make_classification()

nca.fit(X, y)

print(nca.transform(X).shape)

knn = KNeighborsClassifier(metric=nca.get_metric())

knn.fit(X, y) 

print(knn.predict(X[0 : 2, :]))

 

#http://contrib.scikit-learn.org/metric-learn/generated/metric_learn.NCA.html#metric_learn.NCA