如何创建多层感知器网络实例以用于装袋分类器？

Question

i am trying to create an instance of multi-layer perceptron network to use in bagging classifier. But i don't understand how to fix them.

Here is my code:

My task is:

1-To apply bagging classifier (with or without replacement) with eight base classifiers created at the previous step.


It would be really great if you show me how can i implement this to my algorithm. I did my search but i couldn't find a way to do that

Answer 1

To train your BaggingClassifier :

from sklearn.datasets import load_digits
from sklearn.model_selection import train_test_split
from sklearn import preprocessing
from sklearn.model_selection import train_test_split 
from sklearn.preprocessing import StandardScaler  
from sklearn.neural_network import MLPClassifier 
from sklearn.ensemble import BaggingClassifier
from sklearn.metrics import classification_report, confusion_matrix

#Load the digits data:

X,y = load_digits(return_X_y=True)

X_train, X_test, y_train, y_test = train_test_split(X,y, test_size=0.3, random_state=0)
# Feature scaling
scaler = StandardScaler()  
scaler.fit(X_train)
X_train = scaler.transform(X_train)  
X_test = scaler.transform(X_test)
# Finally for the MLP- Multilayer Perceptron
mlp = MLPClassifier(hidden_layer_sizes=(16, 8, 4, 2), max_iter=1001)

clf = BaggingClassifier(mlp, n_estimators=8)
clf.fit(X_train,y_train)

To analyze your output you may try:

y_pred = clf.predict(X_test)
cm = confusion_matrix(y_test, y_pred, labels=clf.classes_)
print(cm)

To see num of correctly predicted instances per class:

print(cm[np.eye(len(clf.classes_)).astype("bool")])

To see percentage of correctly predicted instances per class:

cm[np.eye(len(clf.classes_)).astype("bool")]/cm.sum(1)

To see total accuracy of your algo:

(y_pred==y_test).mean()

EDIT

To access predictions on a per base estimator basis, ie your mlps, you can do:

estimators = clf.estimators_
# print(len(estimators), type(estimators[0]))
preds = []
for base_estimator in estimators:
    preds.append(base_estimator.predict(X_test))