def accuracy_test(t_size, add_noise=False, sigma_sqr=None, ret_acc=True):
acc = 0
for i in range(NUM_RANDOM):
train_data, test_data = load_train_test_data(t_size, True)
if add_noise:
train_data = add_gaussian_noises(train_data, sigma_sqr)
train_data_mat = get_data_matrix(train_data)
x_mean, x_cov = train_params(train_data_mat)
acc += test_stat(x_mean, x_cov, test_data)
acc /= NUM_RANDOM
print('all accuracy: ', acc)
if ret_acc:
return acc
else:
return 1 - acc
def accuracy_test(t_size, add_noise=False, sigma_sqr=None, ret_acc=True):
acc = 0
for i in range(NUM_RANDOM):
train_data, test_data = load_train_test_data(t_size, True)
if add_noise:
train_data = add_gaussian_noises(train_data, sigma_sqr)
df_param, c_centroids = fisher_discriminant_features(train_data[1], train_data[0])
acc += test_stat(euclidean_dist, test_data, df_param, c_centroids)
acc /= NUM_RANDOM
if ret_acc:
return acc
else:
return 1 - acc
def accuracy_test(t_size, hp, add_noise=False, sigma_sqr=None, ret_acc=True):
acc = 0
for i in range(NUM_RANDOM):
train_data, test_data = load_train_test_data(t_size, True)
if add_noise:
train_data = add_gaussian_noises(train_data, sigma_sqr)
train_data_mat = get_data_matrix(train_data)
w = train_w(train_data_mat, hp)
acc += test_stat(w, test_data)
acc /= NUM_RANDOM
if ret_acc:
return acc
else:
return 1 - acc
def accuracy_test(t_size, add_noise=False, sigma_sqr=None, ret_acc=True):
acc = 0
for i in range(NUM_RANDOM):
(train_tgt,
train_feat), (test_tgt,
test_feat) = load_train_test_data(t_size, True)
if add_noise:
(train_tgt, train_feat) = add_gaussian_noises(
(train_tgt, train_feat), sigma_sqr)
acc += test_stat(euclidean_dist, 3, train_feat, train_tgt, test_feat,
test_tgt)
acc /= NUM_RANDOM
if ret_acc:
return acc
else:
return 1 - acc