Created
December 4, 2013 07:34
-
-
Save mrgloom/7783666 to your computer and use it in GitHub Desktop.
MNIST classifier test with default params.
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
import numpy as np | |
from sklearn.svm import SVC | |
from sklearn.svm import LinearSVC | |
from sklearn.linear_model.stochastic_gradient import SGDClassifier | |
from sklearn.datasets import fetch_mldata | |
from sklearn.utils import shuffle | |
import time | |
#out-of-core \ online | |
#http://scikit-learn.org/stable/auto_examples/applications/plot_out_of_core_classification.html | |
#use all digits | |
mnist = fetch_mldata("MNIST original") | |
X_train, y_train = mnist.data[:70000] / 255., mnist.target[:70000] | |
X_train, y_train = shuffle(X_train, y_train) | |
X_test, y_test = X_train[60000:70000], y_train[60000:70000] | |
X_train, y_train = X_train[:60000], y_train[:60000] | |
#SVC() t:1267.66 acc:94,04 | |
#LinearSVC() t:188.17 acc:91,23 | |
#SGD t:6.09 acc:87,19 # we must shuffle data? \ predicting rate fluctuating => depends on order of dataset? | |
#online SGD t:4.64 acc:~86 (max:88.1 stop criteria?) 10 passes => acc:90.8 | |
#test on more data? => better predictions? (with compared time). | |
#test with grid search => time? | |
#test SVC | |
# clf = SVC() | |
# t0 = time.time() | |
# clf.fit(X_train, y_train) | |
# print (time.time()-t0) | |
# score= clf.score(X_test, y_test) | |
# print score | |
# test linearSVM | |
# clf = LinearSVC() | |
# t1 = time.time() | |
# clf.fit(X_train, y_train) | |
# print (time.time()-t1) | |
# score= clf.score(X_test, y_test) | |
# print score | |
# test SGD | |
# clf = SGDClassifier() | |
# t2 = time.time() | |
# clf.fit(X_train, y_train) | |
# print (time.time()-t2) | |
# score= clf.score(X_test, y_test) | |
# print score | |
#test SGD online # need to do more then 1 pass? | |
step =1000 | |
batches= np.arange(0,60000,step) | |
clf = SGDClassifier() | |
all_classes = np.array([0,1,2,3,4,5,6,7,8,9]) | |
t3 = time.time() | |
max_acc= 0 | |
n_pass= 10 | |
for i in range(0, n_pass): | |
for curr in batches: | |
X_curr, y_curr = X_train[curr:curr+step], y_train[curr:curr+step] | |
clf.partial_fit(X_curr, y_curr, classes=all_classes) | |
score= clf.score(X_test, y_test) | |
if(max_acc<score): | |
max_acc= score | |
print score | |
print (time.time()-t3) | |
print max_acc |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment