def test_with_sparse_code(components=np.loadtxt('components_of_convfeat.txt')):
(X_train, y_train), (X_test, y_test) = util.load_feat_vec()
X_train_codes = np.loadtxt('sparse_codes_of_convfeat.txt')
clf = LogisticRegression(penalty='l1', multi_class='ovr')
clf.fit(X_train_codes, y_train)
X_test_codes = sparse_encode(X_test, components)
print "mean accuracy", clf.score(X_test_codes, y_test)
def selecting_non_zero_coef():
"""
No End
"""
(X_train, y_train), (X_test, y_test) = util.load_feat_vec()
print "original X_train shape", X_train.shape
X_new = LinearSVC(C=0.01, penalty="l1", dual=False).fit_transform(X_train, y_train)
print "selected X_new shape", X_new.shape
def main():
n = 1000
(X_train, y_train), (X_test, y_test) = util.load_feat_vec()
digits_proj = TSNE(random_state=130).fit_transform(X_train[:n])
del X_train
del X_test
del y_test
scatter(digits_proj, y_train[:n])
plt.savefig('t-sne_embedding_{}.png'.format(n), dpi=120)
def sparse_logisitic_regression():
"""
If convolutional neural network features are highly overfitting.
Then we could select features from sparse model.
>> mean accuracy 0.352866
"""
(X_train, y_train), (X_test, y_test) = util.load_feat_vec()
clf = LogisticRegression(penalty='l1', multi_class='ovr')
clf.fit(X_train, y_train)
print "mean accuracy", clf.score(X_test, y_test)
