如何实现MLP分类器的随机搜索优化？

Question

我是优化新手，一直在尝试学习随机搜索。在学习了一些理论之后，我尝试在我以前工作过的MLPC分类器中实现它

def hyperparameter_tune(clf, parameters, iterations, X, y):
  randomSearch = RandomizedSearchCV(clf, param_distributions=parameters, n_jobs=-1, n_iter=iterations, cv=6) 
  randomSearch.fit(X,y)
  params = randomSearch.best_params_
  score = randomSearch.best_score_
  return params, score

此方法在对给定数据集执行随机搜索后返回最佳参数集和最佳分数

parameters = {
    'nohn': [150,200,250,300],
    'solver': ['sgd', 'adam', 'lbfgs'],
    'activation': ['relu', 'tanh']
}
clf = MLPClassifier(batch_size=256, verbose=True, early_stopping=True)
parameters_after_tuning, score_after_tuning = hyperparameter_tune(MLPClassifier, parameters, 20, X_train_pca, y);
print(score)

起初，我只是想优化MLPClassizer的隐藏神经元数量、解算器和激活函数。因此，在创建分类器时，我为其他参数（如batch_size）指定了一个固定值。但当我将分类器传递给HyperParameter_tune方法时，我得到以下错误

---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
<ipython-input-18-a0c800c38881> in <module>()
      1 clf = MLPClassifier()
----> 2 parameters_after_tuning, score_after_tuning = tuning(MLPClassifier, parameters, 20, X_train_pca, y);
      3 print(score)

/usr/local/lib/python3.6/dist-packages/sklearn/base.py in clone(estimator, safe)
     65                             "it does not seem to be a scikit-learn estimator "
     66                             "as it does not implement a 'get_params' methods."
---> 67                             % (repr(estimator), type(estimator)))
     68     klass = estimator.__class__
     69     new_object_params = estimator.get_params(deep=False)

TypeError: Cannot clone object '<class 'sklearn.neural_network._multilayer_perceptron.MLPClassifier'>' (type <class 'abc.ABCMeta'>): it does not seem to be a scikit-learn estimator as it does not implement a 'get_params' methods.

有人能帮我吗。在我的代码中，很可能有很多事情我可以做得更好。我很想听到任何建议

Answer 1

你的代码中有不止一个bug

1. 您正在将MLPClassifier类传递给hyperparameter_tune，而不是clf

2. MLP没有nohn参数

3. 您正在打印不存在的score

下面是一个运行的最小示例：

from sklearn.model_selection import RandomizedSearchCV
from sklearn.neural_network import MLPClassifier
import numpy as np

def hyperparameter_tune(clf, parameters, iterations, X, y):
  randomSearch = RandomizedSearchCV(clf, param_distributions=parameters, n_jobs=-1, n_iter=iterations, cv=2) 
  randomSearch.fit(X,y)
  params = randomSearch.best_params_
  score = randomSearch.best_score_
  return params, score


parameters = {
    'solver': ['sgd', 'adam', 'lbfgs'],
    'activation': ['relu', 'tanh']
}
clf = MLPClassifier(batch_size=256, verbose=True, early_stopping=True)

X_train_pca = np.random.randn(6,2)
y = [0, 1, 0, 1, 0, 1]
parameters_after_tuning, score_after_tuning = hyperparameter_tune(clf, parameters, 20, X_train_pca, y);
print(score_after_tuning)