如何用sklearn改变MLP网络的学习率和隐藏层

Question

我正在用sklearn.cross_validation.cross_val_score函数对多层感知器进行交叉验证

from sklearn import svm
import numpy as np
from sklearn.model_selection import cross_val_score
clf = svm.SVC(gamma='auto')
scores = cross_val_score(clf, X, np.ravel(y), cv=5, scoring='accuracy')

交叉验证正在进行，返回0.8579100145137881。如何使用sklearn更改学习率或隐藏层的数量？我想提高准确性。

Answer 1

from sklearn import svm
import numpy as np
from sklearn.neural_network import MLPClassifier
clf = MLPClassifier(solver='sgd', hidden_layer_sizes=(4,4), learning_rate_init=0.05, activation='logistic', max_iter=30000)
from sklearn.model_selection import cross_val_score
scores = cross_val_score(clf, X, np.ravel(y), cv=5, scoring='accuracy')